Health GIS - Enabling Health Spatially

INVESTIGATION OF TUBERCULOSIS CLUSTERS IN DEHRADUN CITY USING GEOGRAPHICAL INFORMATION SYSTEM AND SPATIAL SCAN STATISTIC Neeraj Tiwari *, K.Ram Mohan Rao**, V.S.Tolia*** * Department of Statistics, Kumaun University, S.S.J.Campus, Almora-263601, ** Geoinformatics Division, Indian Institute of Remote Sensing (NRSC), Dehradun 248001, *** District Tuberculosis Officer, Dehradun 248001. [email protected] , [email protected] ABSTRACT: World health organization has declared Tuberculosis a global emergency in1993. It has been estimated that one third of the world population is infected with Mycobacterium tuberculosis, the causative agent of tuberculosis. The emergence of TB/HIV co-infection poses an additional challenge to the control of tuberculosis throughout the world. The World Health Organization is supporting many developing countries to eradicate tuberculosis. It is an agony that one fifth of the tuberculosis patients worldwide are in India. The eradication of tuberculosis is the greatest public health challenge for this developing country. The aim of the present population based study on Mycobacterium tuberculosis is to test a large set of tuberculosis cases for the presence of statistically significant geographical clusters using Geographical Information Systems and Spatial scan statistics. Significant (p < 0.05 for primary clusters and p < 0.1 for secondary clusters) high rate spatial and space-time clusters were identified in seven wards of the Dehradun Municipal area. There is sufficient evidence about the existence of statistically significant tuberculosis clusters in seven wards of Dehradun Municipal area of Uttarakhand, India. KEY WORDS: Geographical Information System, Mycobacterium tuberculosis, Spatial scan statistics, TB/HIV. 1. INTRODUCTION prevalence of TB and HIV co-infection worldwide is 0.18% and about 8% TB cases have HIV infection [3]. 1.1 Background According to the WHO Report 2004 on Global TB Con- Tuberculosis (TB) is an infectious disease caused by the trol, India is sharing 20% burden of TB patient world- bacillus Mycobacterium tuberculosis and spreads through wide, and is leading the 22 high burden countries in the air by a person suffering from TB. The 1990 World world. The estimated number of TB cases in India is Health Organization (WHO) report on the Global Burden 422.6 million (44% of the total population) with 1.8 mil- of Disease ranked TB as the seventh most morbidity- lion people developing TB every year and nearly 0.5 mil- causing disease in the world, and expected it to continue lion dying annually due to TB [7]. More than 1000 peo- in the same position up to 2020 [1]. In 2001, the WHO ple in a day and one in every minute die of TB in India estimated that 1.86 billion persons (32% of the world [8]. In India, Every year TB results in 300,000 children population) were infected with TB. Each year, 8.74 leaving schools. Economic burden to the society is to the million people develop TB and nearly 2 million die. This tune of approximately $ 3 billion [4]. The emergence of means that someone somewhere contracts TB every four TB/HIV co-infection poses an additional challenge to the seconds and one of them dies every 10 seconds [2,3]. control of TB in India. According to the 2006 Report on Unless properly treated, an infectious pulminory TB (i.e., Global AIDS Epidemic, India now has the highest num- the TB of lungs) patient can infect 10-15 people in a year ber of people living with HIV. It has 5.7 million HIV in- [4]. TB is the most common opportunistic disease that fected persons by the end of 2005, against 5.5 million in affects people infected with HIV. As HIV debilitates the South Africa [9]. immune system, vulnerability of TB is increased many fold. It is estimated that without HIV, the lifetime risk of With such a magnitude of disease and looming danger of TB-infected people developing tuberculosis is only 10%, HIV co-infection, TB is the biggest public health compared to over 50% in the case of people co-infected challenge for India. Proportion of deaths in 18-59 year with HIV and TB [5,6]. HIV is also the most powerful age group is 80% [8]. TB kills more women than all risk factor for the progression of TB-infection to the causes of maternal mortality combined. It also adversely disease. In a reciprocal manner, TB accelerates the affects child-care. A substantial proportion of female progression of HIV in to AIDS (Acquired Immune infertility cases are also caused by TB. The public sector Deficiency Syndrome), thus shortening the survival of health expenditure is 0.9% of GDP and less than 10% of patients with HIV infection. Fortunately, TB is a curable Indians have access to any health insurance [10]. The disease even among the HIV-infected people. The current annual per capita public health expenditure in the country is no more than Rs. 200 [11].

To contain this scourge, the National Tuberculosis Con- coordinate system can then be layered together for trol Program (NTCP) was adopted in India in 1962. How- mapping and analysis. ever, the desired results were not forthcoming. There was over dependence on X-ray for diagnosis. Treatment regi- 1.3 Disease Control mens used were often non-standard and incomplete treat- ment was the norm rather than an exception. On recom- The use of GIS with spatial statistics, including spatial mendations of an expert committee in 1992, a revised filtering and cluster analysis has been applied to many strategy to control TB was adopted under the name of Re- diseases to analyze and more clearly display the spatial vised National Tuberculosis Control Program (RNTCP). patterns of disease [15]. Spatial scan statistic implemented in SaTScan software was successfully used The basic problems in geographical surveillance for a to detect the clusters of different diseases worldwide. spatially distributed disease are the identification of areas However, the scan statistic is used for the first time in the of exceptionally high prevalence, to test their statistical present study to detect the clusters of Mycobacterium significance and to identify the reasons behind the tuberculosis. A 30-month (January 1994-June 1996), elevated prevalence of the disease. A hotspot is an area of prospective, city-wide study of all cases of TB using high response or an elevated cluster for an event. traditional contact investigations, GIS data, and Temporal, spatial and space-time scan statistics are molecular epidemiological comparison of commonly used for disease cluster detection and Mycobacterium tuberculosis was carried out at Baltimore, evaluation. Some of them are either able to detect clusters in which clusters of recently transmitted cases of TB were with no inference involved, or they do inference without detected in geographically distinct areas of Baltimore the ability to detect the location of clusters. However, the [16]. A population-based cross-sectional study on all spatial scan statistic developed by Martin Kulldorff [12] incident culture-positive TB cases reported in New Jersey can both detect and provide inference for spatial and from January 1996 to September 1998 using multiple space-time disease clusters. The spatial scan statistic molecular techniques in conjunction with surveillance implemented in SaTScan software [13] offers several data was used to identify a previously unidentified advantages over the existing techniques for detection of outbreak of Mycobacterium tuberculosis in a defined disease clusters. Temporal, spatial and space-time scan geographical setting [17]. Transmission of TB during a statistics [14] are now commonly used for disease cluster nine year period (1988-1996) in a countrywide detection and evaluation, for many diseases including community-based cohort of HIV-infected persons is cancer, Creutzfeldt-Jakob disease, granulocytic Switzerland was investigated and it was found that one ehrlichiosis, sclerosis, diabetes, and giardiasis. There had fourth of TB cases were grouped in clusters of 138 HIV- been no studies to detect the statistically significant infected patients. A study was conducted in the clusters of TB in Uttaranchal, India. The detection of municipality of São Paulo, Brazil, from 1994 to 1998 to these clusters may be highly useful in surveillance of the describe the distribution of TB mortality by area and to disease and finding the factors behind the spread of the evaluate its statistical association with several population disease and make suitable policies to control these characteristics. factors. 2. STUDY AREA 1.2 Basics of Geographical Information System 2.1 Location and Situation GIS refers to a system of hardware and software used for storage, retrieval, mapping, analysis and integration of The city of Dehradun is situated in the south central part geographical and non-geographical features. The of the Dehradun district of Uttarakhand state. Dehradun durability and platform independency of data for city lies at 30o19’ N and 78o20’ E. The area under the developmental planning has proved inadequate in the administrative control of the Dehradun Municipal Board traditional methods of paper based data handling. The is 38.04 sq. km. The Dehradun Municipal Board is GIS provides a systematic spatial and attribute database, divided into 45 wards according to 2001 Census. Two which is prerequisite for implementing development and intermittent streams viz. Rispana River and Bindal River, research projects for more timely response to promote on the east and west respectively, mark the physical limit effective administration, planning, decision-making and of Dehradun Municipality. For the control of tuberculosis development process. Digital database can be used as an in the region, the Dehradun district has been divided in to input for E-governance. It facilitates drawing three Tuberculosis Units (TUs), situated at Dehradun, developmental strategies that are sustainable, area Rishikesh and Chakrata. The microscopic test of sputum specific and takes into account the local needs. Spatial for detection of tuberculosis can be carried out at sixteen features are stored in a coordinate system Microscopic Centres (MCs) of the district. These MCs (latitude/longitude, state plane, UTM, etc.), which are situated at Dehradun, Premnagar, Sahaspur, references a particular place on the earth. Descriptive Vikasnagar, Raphael Home, St. Paul, Christain Hospital attributes in tabular form are associated with spatial Harbetpur, Rishikesh, Doiwala, Jolly Grant, Raipur, features. Spatial data and associated attributes in the same Chakrata, Sahiya, Kalsi and Tuni.

3. METHODOLOGY The maximum spatial cluster size was first set to in- clude up to 50% of population for both excesses and 3.1 Data Collection deficits and then set at 10% and 5%, to test the ex- cesses and deficits separately because testing at the For the present study, we have collected the secondary 10% and 5% levels can identify smaller, more de- data for the tuberculosis patients treated in Dehradun fined areas. For statistical inference, 999 Monte Tuberculosis Unit of Dehradun district, situated at Doon Carlo replications were performed. For purely spatial hospital Dehradun. The data was recorded directly from analysis, the null hypothesis of no significant clusters the Tuberculosis Register of this TU during January 1, is to be rejected when the simulated p-value was less 2007-May 31, 2008. This data contained the personal than or equal to 0.05 for the primary clusters and 0.1 information like age, sex, religion, address, etc. of the for the secondary clusters since the later have conser- patient and also regarding his/her treatment details. On vative p-values. For geographical analysis we have the basis of these details, we had divided the patients into used the techniques available through the Geographi- 45 wards of Dehradun Municipal area created on the cal Information System (GIS). All the geographical basis of 2001 census. The list of these wards and their and cartographic outputs have been presented using population and area details were taken from Dehradun the ArcGIS 9 application software. Municipal Board. During January 1, 2008-May 31, 2008, there were in all 829 TB patients registered in dehradun 4. RESULTS AND DISCUSSIONS TU, out of which 478 patients (302 male and 176 female patients) belonged to Dehradun Municipal area. Using the maximum spatial cluster size of ≤5% of the to- Similarly, during January 1, 2007-December 31, 2007, tal population, the spatial cluster analysis identified the there were in all 1850 TB patients registered in dehradun most likely significant cluster for high occurrence of TB TU, out of which 1044 patients (659 male and 385 female in the Dharampur ward of Dehradun for 2008. The over- patients) belonged to Dehradun Municipal area. all relative risk (RR) within the cluster was 2.806 with an observed number of 29 cases treated between January 1, 3.2 Technologies being used 2008 and May 31, 2008, compared with 10.75 expected cases. Statistically significant secondary cluster for high The spatial scan statistic developed by Kulldorff im- occurrence of TB were also detected at Khudbura, plemented in a software program, SaTScan v6.1, is Ajabpur, Adhoiwala and Patel Nagar wards. The details used to test the presence of statistically significant of these clusters are provided in Table 1. spatial clusters of TB and to identify their approxi- mate locations. Purely spatial analysis, which does Table 1 High Rate Clusters for 2008 not take time into account, was performed to detect the TB clusters in the study region. The theory be- Location No. of Rela- Log P- hind the spatial scan statistic is a generalization of a cases tive Likelih Value test proposed by Turnbull et al.. The number of risk hood events may be assumed to be either Poisson or Dharampur 29 Ratio 0.001 Bernoulli distributed, depending on the application Khudbura 25 2.806 0.001 of the data under study. The spatial scan statistic im- Ajabpur 35 2.945 10.88 0.022 poses a circular window on the map and lets the cen- Adhoiwala 33 1.852 10.19 0.039 ter of the circle move over the area so that at differ- Patel Na- 18 1.845 0.057 ent positions the window includes different sets of gar 2.262 5.19 neighboring census areas. If the window contains the 4.85 centroid of the census area, then that whole area is included in the window. For each circle centroid, the 4.54 radius of the circular window is varied continuously from 0 up to a maximum radius so that the window Using the maximum spatial cluster size of ≤5% of the to- never includes more than 50% of the total population tal population, the spatial cluster analysis identified the at risk. The spatial scan statistic is based on the like- most likely significant cluster for high occurrence of TB lihood ratio test. As the likelihood ratio is maximized in the Adhoiwala ward of Dehradun for 2007. The overall over all the circles, it identifies the circle that consti- relative risk (RR) within the cluster was 3.116 with an tutes the most likely cluster. Its p-value is obtained observed number of 117 cases treated between January 1, through Monte Carlo hypothesis testing technique 2007 and December 31, 2007, compared with 40.62 ex- proposed by Dwass. To find the distribution of the pected cases. Statistically significant secondary cluster test statistic, 999 random Monte Carlo replicates of for high occurrence of TB were also detected at Dharam- the data set under the null hypothesis of no signifi- pur, Gandhi Gram, Rajpur, Ajabpur and Khudbura. The cant clusters are generated, calculating the test statis- details of these clusters are provided in Table 2. tic for each replica. Table 2 High Rate Clusters for 2007 Identification of spatial high/low clusters was done under Poisson probability model assumption. Location No. Relative Log P-

of Risk Likeli Value Karanpur, 13 0.328 11.83 0.001 Cases hood Ra- Bakralwala 02 0.001 tio 0.001 15 0.110 11.67 0.003 Adhoiwala 117 3.116 0.001 Shivaji 0.399 8.56 Dharampur 64 2.817 50.35 0.001 23 0.005 Gandhi 60 2.148 Mansingh Gram 24.17 wala, 09 0.022 Rajpur 57 2.161 Rispana 09 0.071 Ajabpur 77 1.854 13.30 07 0.084 Khudbura 39 2.056 Idgah, Sridev 0.472 8.13 Suman Na- 12.84 0.001 gar 0.376 6.040 11.44 0.003 0.417 4.642 7.88 0.008 Bhandarib- 0.380 4.593 agh The details about the low rate tuberculosis clusters for the years 2008 and 2007 are provided in Tables 3 and 4, re- Race spectively. Course(N) Salawala Table 6 provides the information regarding the high rate zone of tuberculosis for the year 2008. Figure 1 High rate TB clusters in 2007 Figure 1 shows the spatial representation of hhigh rate Figure 2 High rate TB zones in 2008 clusters in 2007. Table 3 Low Rate Clusters for 2008 Location No. Rela- Log P- It consists of 19 wards of the city with total number of of tive Likeli- Value 274 TB cases reported between January to May 2008. Mansingh- Cases Risk hood Figure 2 shows the geographical view of high rate TB wala 0 0 0.004 zones in 2008. Arya nagar 8.488529 Maha rani- 01 0.084 Table 5 High Rate Zone for 2008 bag 02 0.147 Salawala 8.306177 0.005 Location No. Rela- Log P- 01 0.119 7.646203 0.010 (Ward of tive Likeli- Value No.) Cases Risk hood 5.234072 0.065 0.001 14, 38, 28, 274 1.655 15.04 37, 29, 30, Table 4 Low Rate Clusters for 2007 39, 13, 16, 12, 15, 27, Location No. Rela- Log P- 26, 32, 36, of tive Likeli Value 17, 10, 31, Maharan cases Risk hood 33. Ibag 25.25 0.001 Arya 01 0.033 Nagar 12.63 0.001 Jhanda 05 0.192 Table 6 provides the information regarding the low rate Mohalla 12.39 0.001 zone of tuberculosis for the year 2008. It consists of 22 02 0.105 wards in 5 groups with 77, 86, 5, 28 and 15 TB cases, re- spectively, reported between January to May 2008.

Table 6 Low Rate Zone for 2008 epidemiological situation of tuberculosis in the Dehradun city of Uttarakhand, India. This will serve as a baseline Location No. Rela- Log P- for evaluating the impact of disease control measures and (Ward No.) of tive Likeli Value epidemiological trends in the coming years. This type of Cases Risk hood studies can help the district tuberculosis units to identify 20, 23, 22, 21, 77 0.408 0.001 areas of high tuberculosis prevalence and chalk out 19, 24, 25, 43 37.06 strategies in a more focused way. This would initiate 40, 42, 35, 36, 86 0.440 intensified case finding activities, further promotion of 7, 34, 41, 31, 33.4366 0.001 general health and hygiene, improving nutritional status 39 05 0.192 96 of the community, compulsory BCG immunization of the 3 children, and better coordination of government and 12.6346 0.001 private sector in the hotspots detected by the study. The 14, 38 28 0.478 40 scope of present study is limited to only the capital city of 5,6 15 0.399 0.002 Uttarakhand. There is a scope to involve the whole state 9.47 0.004 of Uttarakhand and then the country as a whole. This 8.56 would mean a critical appraisal of the RNTCP in the whole country. Further, it has been found that the Cluster analysis identifies whether geographically hospitals of good repute get sizeable number of patients grouped cases of disease can be explained by chance or from other tuberculosis units (TU). Therefore, District are statistically significant. It detects true clusters of Tuberculosis Centres (DTC) must work to activate all disease from cases grouped around population canters. hospitals by posting adequate staff and providing better Figure 3 shows high rate TB clusters in 2008. facilities over there. One limitation of DOTS program is that it is more hospital centric, laying more emphasis on diagnosis and treatment and less on case finding at community level. Our study can identify areas of high prevalence, leading to intensification of case finding activities by district tuberculosis control units. This strategy can be highly useful in eradication of TB from the country. Future research will include investigating the effect of various socio-economic and environmental factors on the occurrence of TB in the hilly region of the state. Figure 3: High rate TB cluster in 2008 REFERENCES The results of the study suggest that there are statistically 1. Murray CJL, and Lopez AD: The global burden of significant hotspots of Mycobacterium tuberculosis in disease: a comprehensive        assessment of mortality five wards for 2008 and six wards of the city for 2007. and disability from diseases, injuries and risk factors One high rate zone of tuberculosis was also detected in in 1990 and projected to 2020.World Health Organization Dehradun city, consisting of 19 wards for the year 2008. Document 1996, W 74       96GL - 1/1996. The use of GIS with spatial statistics, including spatial filtering and cluster analysis has been applied to many 2. Narain JP (ed.): Tuberculosis-epidemiology and diseases to analyze and more clearly display the spatial control. World Health Organization, Regional Office for patterns of disease [28-33]. Spatial scan statistic [12] South East Asia, New Delhi, India 2002, implemented in SaTScan software [13] was successfully SEA/TB/2002. 248:15-18. used to detect the clusters of different diseases worldwide [17-27]. However, the scan statistic is used for the first 3. Dye C, Scheele S, Dolin P, Pathania V and Raviglione time in the present study to detect the clusters of MC: Global burden of       disease: estimated incidence, Mycobacterium tuberculosis. prevalence, and mortality by country. J Am Med Assoc 1999, 282: 677-686. 5. CONCLUSIONS 4.TB India 2005. RNTC Status Report. Central TB In our study, significant clusters were identified in seven Division, Directorate General of        Health Services, wards of Dehradun city. The results of the present study New Delhi. [http://www.tbcindia.org]. provide useful information on the prevailing 5. Cauthen GM, Pio A, and ten Dam HG: Annual risk of infection. World Health        Organization Document 1988, WHO/TB/88.154: 1-34.

6.Telzak EE: Tuberculosis and Human Immunodeficiency Virus infection. Med  Clin North Am 1997, 81: 345-360. 7.World Health Organization. Global Tuberculosis Control-WHOReport 04,         WHO/HTM/TB/2004.331. 8.TBC India Template. [http://tbcindia.org/template.htm]. 9.UNAIDS. Report on global AIDS epidemic 2006, UN- AIDS/06.20E. 10. Census of India 2001. Registrar General and Census Commissioner, India.          [http:// www.censusindia.net] . 11. National Health Policy 2002. Ministry of Health and Family Welfare, New Delhi,         India. [http://mo- hfw.nic.in/np2002.htm]. 12. Kulldorff M: A spatial scan statistic. Communica- tions in Statistics: Theory and               Methods 1997, 26:1481-1496. 13. Kulldorff M and Information Management Services, Inc. SaTScanTM v6.1: Software        for the spatial and space-time scan statistics. http://www.satscan.org/, 2006. 14. Wallenstein S: A test for detection of clustering over time. Am J Epidemiol 1980,         111: 367-372. 15. Curtis A: Using a spatial filter and a geographic information system to improve rabies surveillance data. Emerg Infect Dis 1999, 5: 603-606. 16. Bishai WR, Graham NMH, Harrington S, Pope DS, Hooper N, Astemborski J, Sheely L, Vlahov D, Glass GE, Chaisson RE: Molecular and geographic patterns of tuberculosis transmission after 15 years of Directly Observed Therapy. J Am Med Ass 1998, 280, 1679-1684. 17. Bifani PJ, Mathema B, Liu Z, Moghazeh SL, Shopsin B, Tempalski B, Driscoll J, Frothingham R, Musser JM, Alcabes P, Kreiswirth BN: Identification of a W variant outbreak of Mycobacterium tuberculosis via population based molecular epidemiology. J Am Med Ass 1999, 282, 2321-2327.

Technical Session - 6 Telemedicine and Advance Health Instruments Diagnosing Arrhythmia of Patients by Detecting R-R Intervals in ECG Signal Shilpa S Joshi and P. T. Karule……………………………..……………………………………….……137 Telemedicine and Location Based Services for Medical Operations Madhav Prashanth Ramachandran …………………………………………………………………….…141 Twitter in Healthcare – A Boon or a Bubble? Sivaram Inguva and Jayakanth Dornadula……………………………………………………….………145 Critical Evaluation of Software Based Videoconference Solution for Telemedicine Repu Daman Chand and S K Mishra………………………………………………………………………146 Telemedicine Secure Data Hybrid Network Intrusion Detection Model B. Naga Malleswara Rao, N. Sambasiva Rao and V. Khanaa……………………………………………147

DIAGNOSING ARRHYTHMIA OF PATIENTS BY DETECTING R-R INTERVALS IN ECG SIGNAL Shilpa S Joshi 1 , P.T.Karule 2 1 Yashwantrao Chavan College of Engg. Nagpur, India [email protected] 2 H.O.D. & Asst. Prof Electronics Dept, Yashwantrao Chavan College of Engg. Nagpur [email protected] ABSTRACT: In this paper, an automatic detecting algorithm for peak detection is applied for analyzing ECG recording and criteria for a dangerous arrhythmia are applied for a type of arrhythmia. In this investigation, sampled ECG recordings from MIT-BIH database are collected for off-line analysis. A combinative application of digital filters for bettering ECG signals and promoting detecting rate for R peak is proposed as pre-processing. The filters are applied to eliminate different types of noises mixed with ECG recordings. The R waves of ECG are detected using slope detection technique and proper thresholding. This algorithm is developed to calculate Heart Rate (HR). HR time series is constructed from the R-peaks. Application of this technique to the 21 sets of ECG data of MIT-BIH database shows the average detection rate is around 95%. Finally the criteria are given for arrhythmia detection. KEY WORDS: Signal processing, electrocardiography (ECG), Heart Rate, arrhythmia. 1. INTRODUCTION treated by digital filters. Those digital filters are supported in toolbox of Matlab. The ECG (Electrocardiogram) signal is a recording of the 2) Peak detection- R-waves are detected by slope heart’s contractile activity in electrical potential and it is detection technique. Before the occurrence of an R-wave, the most commonly used biomedical signal for the the slope is positive and after the R-wave, the slope is detection of asymptomatic arrhythmia and diagnosis of negative. cardiovascular diseases or abnormalities. A disturbance Many QRS detection algorithms [8, 9, 10, and 11] are in the conduction of excitation from the atria to the proposed in succession. Prior to the calculation of ventricles is revealed by the prolongation of the P-R Instantaneous Heart Rate (IHR) we apply de-noising to intervals. Any electrocardiographic lead, which shows a the ECG signal. P and QRS wave, can be used to diagnose 3) Calculation of Heart Rate- To predicts abnormalities atrioventricular conduction defects. The rapid and through R-R interval. objective measurement of timing intervals of the ECG by automated systems is superior to the subjective 2. METHODS AND MATERIALS assessment of ECG morphology. The timing interval measurements are usually made from the onset to the In this investigation, samples of ECG recordings of MIT- termination of any component of the ECG, after accurate BIH database were used for study cases. detection of the QRS complex. Electrocardiogram (ECG) signal: An automated diagnostic system is required to speed up the diagnostic process and assist the cardiologists in ECG signal is a time-series changing of voltage in a examining patients using non-invasive techniques. Such cardio-system. Because ionic currents are the source of system would require innovative signal processing, an ECG signal, it can reflect and record the activity of analysis and classification techniques. heart. The complete waveform in Fig.1 is called an ECG signal, with labels P, Q, R, S and T indicating its The whole process is divided into three parts. They are: distinctive features [4, 13]. 1) Noise elimination – ECG recording are mixtures of QRS complexes, P-wave, T-wave, natural noises, The P-wave arises from depolarization of the atrium. The interference of power sources, interferences from other QRS complex arises from depolarization of the bio-signals, unidentified noises, etc. Digital filters and ventricles. The magnitude of the R-wave within this signal smoothing techniques [1, 3, 6, 7] are common complex is approximately 1 mv. The T-wave arises from techniques applied for purifying ECG recordings. In this repolarization of the ventricle muscle. QRS complex is study ECG recordings from MIT-BIH database are the most significant wave of an ECG signal and R-R interval used for obtaining heart rate.

Fig.1) Schematic of an ECG signal 2) Peak detection- Implementation: R-waves are detected by slope detection technique, because R-wave is the highest peak, it is easy to trace 1) Toolbox of digital filters in Matlab [12]. The significant negative peak of Q-wave closely There are many functions of digital filters in the before to R-wave and the next negative point for S- toolbox MATLAB. In this study, referring the wave after the R-wave then can also be detected. characteristics of ECG data and noises, infinite impulse register (IIR) is applied as digital filter for 3) Calculation of Heart Rate filtering ECG recordings. This digital filter can support all the functions of low pass filter, high pass filter, R peak detection is important in all kinds of ECG band pass filter and band stop filter. It is easy to signal processing. Use of local extreme values is basis modify the function of filter by setting characteristics for QRS detection. The R wave is most often used in parameters of digital filter. calculation of heart period (because of its relatively For the purposes to eliminate the noises and to better large size).The term \"heart rate\" normally refers to the ECG signals, some parameters should be set with rate of ventricular contractions. Ventricular rate can be proper values. The order of filtering is one of those determined by measuring the time intervals between parameters. A higher order of filtering could make the QRS complexes, which is done by looking at the filter to cut off more signals, not just the noises. Here R-R intervals. the second order of filtering is suggested by experiences. Another critical parameter of digital filter Heart rate = 60 / R-R interval -------- (I) is type of filter. In the following section the most common abnormalities based on inspection of sinus rhythms are a) High-pass filter described. The baseline shifting is due to a very low frequency The first thing to examine while analyzing an ECG signal. This very low frequency component is filtered signal is sinus rhythm. out by high pass filtering. The criterion for a Normal Sinus Rhythms is: b) Band-stop filter  A QRS width should be of 0.04 to 0.12 The frequency of band-pass filter is not a single value seconds and be preceded by a P-wave. but a span. As a digital filter in toolbox of Matlab, band-stop filter is a sub-type of band pass filter with an  The rate for a normal sinus rhythm is 60 to action code of ‘STOP’. The function of this digital 100 beats a minute. filter is focus on the purpose of eliminating power-line interference. If the rate is below 60 beats a minute but the rest is the A band pass filtering of the ECG for noise reduction is same it is a Sinus Bradycardia. If the rate is between done by cascading the following filters. 100 to 150 beats a minute with the same intervals it is i) Low pass (IIR filters with 3db frequency cut off ~ a Sinus Tachycardia. When the Pattern becomes 11 Hz). When signals pass through a low pass filter, irregular with normal intervals it is a Sinus the part of signal with frequencies higher than cut off Arrhythmia. frequency will be filtered. The quantitative criteria for arrhythmia diagnosing ii) High pass (IIR filter with 3 db frequency cut off ~ The ill-condition of extreme bradycardia means that 5 Hz). heart beats too slowly. As a quantitative criterion, the IIR filters are the best choice because they give better inequality can be expressed as: performance with lower filter order. The advantage of Heart rate < 40 beats/min or IIR filters is that they have very small coefficient R-R interval > 1.5 sec. compared with FIR filters. The ill condition of tachycardia means beats too fast. As a quantitative criterion, the inequality can be expressed as: Heart rate>100 beats/min or R-R interval<0.6sec 3. RESULTS We analyzed 11 sets of 1-minute duration ECG data of MIT-BIH databases [15].The noisy signal Fig.2] a) and filtering of ECG signals with serious high frequency noise, baseline drift is shown in Fig 2] b). First the original signal from MIT-BIH database is filtered out and then R spikes are detected for that signal. The respective R-R interval is calculated. This can be utilized to calculate heart rate from which abnormalities can be predicted.

S.N. Data (N1)Actual beats (N2)Detected beats 100%Rate of detection= (N2/N1)* min)(Beats/HR 1 16265 95 94 98.94 93.069 2 16272 62 64 103.22 63.829 3 16273 94 93 98.94 91.547 4 16420 95 94 98.94 93.266 5 16483 97 96 98.96 94.475 6 16773 75 75 100 74.081 7 16795 65 70 107.69 69.462 8 16786 70 70 100 69.462 9 16539 79 78 98.7 77.5 10 17052 70 69 98.57 68.352 11 18184 89 88 98.87 86.557 sample value 3 original signal 0.8 R peak 2.5 0.6 200 400 600 800 1000 1200 200 400 600 800 1000 1200 0.4 2 sample number 0.2 1.5 0 1 -0.2 0.5 -0.4 -0.6 0 -0.5 0 -1 0 0.8 Filtered signal Fig 2] a) Original signals with noise. b) Filtered 0.6 Signal. c) R- peaks in the signal 0.4 200 400 600 800 1000 1200 sample value 0.2 sample number 0 S.N. (Beats/HR -0.2 Data -0.4 (N1)Actual beats -0.6 (N2)Detected beats 100%Rate of detection= 0 (N2/N1)* min) 1 Cu03 120 123 97.56 124 2 Cu05 94 100 106.3 100.9 3 Cu06 94 92 97.87 93.7 4 Cu07 115 114 98.94 115.5 5 Cu10 99 103 104 104 6 Cu11 84 82 97.61 83.6 7 Cu12 82 89 108.5 89.4 8 Cu13 101 101 100 102.6 9 Cu28 121 120 99.17 121.8 10 Cu32 100 99 99.0 100.5

Table 1] Analysis of 1 minute duration NSR datasets. Table 2] Analysis of 1 minute duration Tachycardia data sets. 4. CONCLUSION In this paper we have presented an algorithm which detects R peaks and the duration between them. The analysis has been described to provide precise information of timing intervals that define the morphology of the ECG. Noise generally encountered in the clinical environment is automatically eliminated due to inherent characteristics of the pre- processing technique. The best way for reducing the noises in ECG recordings is to figure out the sorts of noises and filter them as possible as we can. In this study an automatic arrhythmia diagnosing application is built up. A fundamental architecture of automatic arrhythmia diagnosing consists of signal filtering, automatic ECG signal detecting and criteria modeling for arrhythmia diagnosing. But for an advanced application, more information hidden in ECG signals should also be detected. The further information includes P-R interval, S-T interval, Q-T interval and distribution of P-waves and T-waves. In this study, the goal is narrowed down for detecting two common ill-conditions of arrhythmia. REFERENCES [1] T. Nguyen G. Strang. Wavelets and Filter Banks. Wellesley- Cambridge Press, 1996 [2] G. Oppenheim J. M. Poggi M. Misiti, Y. Misiti. Wavelet Toolbox. The Math Works , Inc., Natick, Massachusetts 01760, April 2001 [3] S.Haykin, Adaptive Filter Theory, 4thed. Englewood Cliffs, N J :Prentice Hall PTR,2002 [4] G.B. Moody,R.G. Mark and A. L. Goldberger, “Physionet: A web-based resource for the study of physiologic signals”, IEEE Engineering in Medicine and Biology Magazine, vol.20, no.3, pp.70-75,2001 [5] A. Menrad et al. ,” Dual microprocessor system for cardiovascular data acquisition, processing and recording ,” Proc. 1981 IEEE int. conf. Elect. Contr. Instrument, pp. 64-69 [6] V. X. Afonso, W . J. Tompkins, T. Q. Nguyen, K. Michler, and S. Luo , ”Comparing stress ECG enhancement algorithms”, IEEE Engineering in Medicine and Biology Magazine, vol.15, no.3, pp.37-44,1996. [7] N. V. Thakor and Y. S. Zhu ,”Applications of adaptive filtering to ECG analysis: Noise cancellation and arrhythmia detection”,IEEE Trans. Biomedical Engineering, vol.38, no.8, pp.785-794,1991 [8] E. Skordalakis, “Syntactic ECG processing: A review”, Pattern Recognition.,977-985, 1979.vol. 19, pp. 305- 313,1986 [9] Tompkins WJ, Pan J” A real time QRS detection algorithm”, IEEE Trans.on Biomedical Engg.,. BME-32, No. 3:230-235, March 1985. [10] Y. F. Wu and R. M. Rangayyan,”An algorithm for evaluating the performance of adaptive filters for the removal of artifacts in ECG signals ”,in Proc.20th Canadian Conf. Electrical and Computer Engineering (CCECE’07), Vancouver, BC,Canda,2007. [11] R. Polikar. Wavelet tutorial. eBook, March 1999. http://users.rowan.edu. [12] C. Meyer, J. F. Gavela and Harris, ”Combining algorithms in automatic detection of QRS complexes in ECG signals”, IEEE Trans. Information Technology in Biomedicine, vol.10, no.3,pp.468-475,2006.

[13] Rissam HS; Kishore S; Srivastava S; Bhatia ML; Trehan N, “Evaluation of cardiac symptoms by trans-telephonic electrocardiographic (TTEM)monitoring”: preliminary experience. Indian Heart J 50(1):55-8, Jan-Feb 1998. [14] G.M.Friesen,T.C.Jannett,M.A.Jadallash,S.L.Yates, S.R.Ouint and H.T.Nagle,”A Comparison of the Noise Sensitivity of Nine QRS Detection Algorithms”,IEEE Trans on Biomedical Engg,371:85-98 [15] http://www.physionet.org.physiobank/database/ mitdb/

TELEMEDICINE AND LOCATION BASED SERVICES FOR MEDICAL OPERATIONS Madhav Prashanth Ramachandran A/6 Kasyap Apartments, Nehru Nagar, Fourth Street, Adyar, Chennai-600020 Email: [email protected] ABSTRACT: Telemedicine is the ability to provide healthcare services to the common person using telecommunications. Telemedicine integrated with a Location Based Service can be proved a helpful tool in medical operations and diagnosis in the future. The paper lays emphasis on the development of a ‘telemedico-geographic database management system’ called Medicinal and Location Information for Disease Outbreak (MLIDO) and a shortest path algorithm called Heuristic Instinctive Dijkstra’s Request based Algorithm (HIDRA). The MLIDO is essentially web based decision- making software that can be widely used in emergency medicinal diagnosis. The HIDRA is a heuristic algorithm and an implementation or an extension of the Dijkstra’s Algorithm based on factors affecting transportation like traffic, accidents, tree-fall etc. The proposed algorithm is heuristic in nature and can be implemented in the near future for emergency or non-emergency operations. Location Based Services (LBS) can make full use of the proposed algorithm i.e. HIDRA. The MLIDO is a simple yet effective geographic database management system, which consists of many schemas relating a patient, his disease, symptoms and causes, family medical history and problems etc. These two systems, when integrated together, can be proved a satisfactory and effective method for medical operations and diagnosis. A satellite image maybe superimposed in a MLIDO for better results. KEY WORDS: Telemedicine, Location Based Services (LBS), MLIDO, Shortest Path, HIDRA 1. INTRODUCTION images/videos are emailed or placed on the server for the specialist’s access. The emergence and occurrence of deadly diseases like cancer and AIDS etc has grown over the past few years. Location Based Services The world has witnessed millions of people battling with these diseases for survival and unfortunately, the diseases A Location Based Service (LBS) is the capability of a have always emerged the victors. Time has always been mobile device and then provide services based on the the constraint in deciding the fate of the affected persons. available location information. They are services that In this paper, I aim to propose the development of an integrate a mobile device’s location with other information management system, which caters to information to provide benefit to a user (Jochen H. telemedicine operations and the development of a Schiller, Agnès Voisard, 2004) Heuristic Instinctive Dijkstra’s Request Algorithm This means that a person asking for information about a (HIDRA) for shortest path calculation for medical landmark is based upon his location in the city. The LBS operations. deliver geographic information between mobile and/or static users via the internet or any other communications 2. EXISTING CONCEPTS network. The ingredients of the LBS are- The paper makes use of the following technologies 1. Location and techniques for performing the required medical 2. Geographic Data operation- 3. Control Center 4. Communications Telemedicine Shortest Path Problem Telemedicine is the ability to provide interactive healthcare through modern technology and The Shortest Path Problem is the problem of finding out telecommunications. Telemedicine allows patients to visit the shortest path between two nodes or vertices such that with physicians live over video for immediate care or the sum of the weights of its constituent edges is capture videos/stills and patient data are stored and sent minimized (Reinhard Diestel, 2005). An example of to physicians for diagnosis and treatment (Anthony finding the shortest path is the distance between two Charles Norris, 2001). It is used when both health points on a map, where vertices represent road segments providers are not available at the same time. The and weights represent the time taken to travel through provider’s voice or dictation on the patient’s history, those segments.

The Dijkstra’s Shortest Path Algorithm The database is designed in such a way that it can be ac- cessed via the internet and the results of queries and ana- The Dijkstra’s Shortest Path Algorithm solves the lysis performed in the database can be sent to a major single source shortest path problem when edges have healthcare unit. It also consists of the placing a satellite non-negative weights. It starts at a source vertex (s), image of the study area on the screen. Some of the grows into a tree (T) and ultimately spans all vertices schemas that can be introduced in the proposed database reachable from S. This algorithm is often used in routing i.e. MLIDO are as follows- and computer networks. Figure 1 The Dijkstra’s Shortest Path Algorithm ‘O’ is the origin and ‘T’ is the target. The shortest path to reach from O to T is O->A->B->E->D->T. 3. PROPOSED SYSTEM Figure 2 Flow Diagram of the Operations involved In this section, I propose the development of a management system called MLIDO, which would be handy as data in telemedicine and an algorithm called HIDRA that would save and minimize the time taken in reaching the main physician. Requirements of the System The following are the requirements of the proposed system, which would be fruitful for carrying out the diagnosis- 1. Video-Conferencing Unit 2. Internet Connection 3. Location Based Service 4. Satellite Imagery (not a necessity) Medicine and Location Information for Disease Outbreak (MLIDO) The MLIDO is a simple yet effective Geographic Database Management System, which caters to telemedicine operations. The chief ingredients of the MLIDO are as follows- 1. Geographic Information i.e. location and posi- tion of an individual in a given study area 2. Medical records and history of a person and his family member

1. The PATIENT schema that consists of Patient 6. Examination of the patient by doctors and prior- ID, Patient Name, Location, Relation ID and the ity assignment. Disease ID. 7. Decision making. 2. The DISEASE schema consisting of the Disease The most important task is to reach the PHU from the ID, Disease Name and the Symptom ID. MHU in the shortest time i.e. as quickly as possible. A diagram is drawn as follows- 3. The SYMPTOM schema consisting of details re- garding the symptoms and cause of the disease. Figure 3 A network diagrams showing the MHU and the PHU 4. The RELATIONSHIP schema that comprises of the relationships between a patient and his fam- In the above drawn figure, the white circles represent ily members. the MHU and the black ones depict the PHU. The arrows represent the directions or “routes” from the MHU to the Similarly, other schemas may be developed when need PHU. The MHU and PHU are now called the secondary arises. The operations to be performed for medical node (N2) and primary node (N1) respectively. With the diagnosis of the patients are best represented by a flow help of these coordinates, the patient can identify his chart or a flow diagram. The overall process consists of – position and discard the remaining ones. Now, the target the steps shoed in figure 2. reduces to reach anyone of the PHU from a single MHU. The network diagram is shown below- The algorithm or the flow diagram, which was discussed in the previous section, is now elucidated. The Figure 4 A network diagram showing one MHU and the paper considers the cases of normal as well as outbreak other PHUs. cases and provides an efficient way of making effective decisions. The “clinics” are labelled as minor or secondary healthcare units (MHU) and are represented by the letter ‘S’ in the diagrams drawn henceforth. The letter ‘P’ represents major healthcare units i.e. hospitals are referred to as Primary Healthcare Units (PHU). If the medical facilities of the MHU are poor, then a computer having the MLIDO and an internet connection will suffice. The patient’s medical records and history are retrieved by the database as mentioned previously. All it needs is a ‘unique ID card’, something like the Social Security Number (SSN) or a SIM card. The physician at the MHU examines the patient, stores, and updates his problem i.e. infection, probable cause and symptoms and family history and medical problems in the MLIDO management package. Telemedicine takes place in such a way that a videoconference link is created between the physician at the MHU and one doctor present in each PHU. Videoconferencing and telemedicine ensure that long distant patients receive quick and accurate diagnosis and consultations no matter how far away the doctor at the PHU is far from the patient. The process consists of the following steps- 1. View and observe the patient’s health and med- ical records. 2. Search for doctors and medical experts using an inbuilt search engine. 3. Use a GPS to obtain the coordinates of the MHU. 4. Video-Conferencing is done between the MHU physician and all the doctors present at the PHUs. 5. Interaction between doctors takes place.

The doctor’s decision as pointed previously is chosen as The primary function of this algorithm is to minimize the PHU where the patient would be diagnosed the distance as well as the time taken in reaching the PHU completely. The algorithm is simple and explained from the MHU such that patient’s diagnosis can be below- completed faster than expected. The flow of operations of the HIDRA consists of the following steps- 1. The MHU doctor chooses the best PHU for dia- gnosis. 1. Identify the MHU on a map, say a digital map. 2. REQUEST for identification of the main road 2. He gives other alternative PHUs as well. A Personal Digital Assistant (PDA) equipped with a using a Location Based Service. Location Based Service (LBS) is beneficial for this 3. REQUEST for number of streets to be travelled purpose. The network model is shown below where the circles in red show the plausible alternatives. to reach the main road network. 4. Assign a street number to every street where the Figure 4 A network diagrams connecting the MHU with the best PHU and alternative PHUs (shown in red) street number maybe a function of distance and A possible map model of the MHU and PHUs with the number of diversions. 5. If there are many streets to be travelled to reach Figure 5 MHU and PHUs represented on a map the main road, then AVOID. If there are lesser The algorithm for finding out the shortest path streets in number, but the distance to reach the main road is more, then USE. between the MHU and the PHU is referred to as a 6. After reaching the main road, REQUEST for the Heuristic Instinctive Dijkstra’s Request-Based location of the main PHU as well as the alternat- Algorithm (HIDRA). The HIDRA is a request based ive PHUs. heuristic Dijkstra’s Algorithm, which acts on “requests” 7. REQUEST for live traffic updates, accidents and and “instincts”. tree falls such that obstacles can be avoided. 8. DECIDE the path to traverse based on the re- quest. 9. REQUEST for bridges such that the time taken to reach the destination is minimum. Also, look for the alternate PHUs as well. 10. If ANY of the PHUs is close to the bridge, choose that. REQUEST for shortest path via the LBS. The LBS should act on the Dijkstra’s Al- gorithm. This algorithm is a cyclical process since at every point of time, traffic updates, accidents, tree falls and other obstacles have to be updated and fed. The algorithm is also Instinctive in nature since the decision is based on the ‘drivers’ instincts and thoughts. 3. CONCLUSIONS Since, this is only a proposed heuristic algorithm, which is a further extension of the Dijkstra’s Algorithm based on certain factors i.e. obstacles, it can be used as an effective method for performing medical operations where time is a crucial factor. The proposed method that consists of the development of the MLIDO and the use of the HIDRA can be very fruitful. REFERENCES 1. Location-based services, 2004, by Jochen H. Schiller, Agnès Voisard 2. Essentials of telemedicine and telecare, 2001 by Anthony Charls Norris 3. Graph theory, 2005, by Reinhard Diestel

Twitter in Healthcare – A Boon or a Bubble? Abstract Sivaram Inguva and The social–networking revolution is slowly and silently sweeping the healthcare Jayakanth Dornadula industry. In addition to the existing Internet technologies and applications, this networking is also changing the end user perspective of accessing, assimilating and transmitting healthcare information. On one hand the consumers and patients are using the networking sites to know about recent and relevant information on medical products, symptoms, treatments and outcomes and on the other medical agencies, hospitals and companies are trying to reach consumers by providing information on products, treatment options, clinical trials, recruitment process and branding using these tools. One such social networking tool that caught the attention of medical and healthcare community is Twitter. Twitter is a free social networking and micro-blogging service that enables its users to send and read posts up to 140 characters. It is perceived that this would bring radical changes by the way of providing an interactive platform to the Healthcare and medical community. The benefits of Twitter include blood glucose tracking, maintaining personal health diary, adverse event reporting, pharmacovigilance, drug safety alerts and so on. However, medical community is skeptical on the quality of data transmitted under HIPAA guidelines. The present article talks about the advantages and limitations of the using social networking tools like twitter.

Critical Evaluation of Software Based Videoconference Solution for Telemedicine Abstract Repu Daman Chand Video-conference is the main component of any telemedicine solution. Currently, and S K Mishra telemedicine platforms deployed across the country are hardware based. For mass scale deployment it will work out expensive. Alternate video-conference solution based on software needs to be worked out which not only meet the purpose but also work out cheaper. A study to this effect was carried out at the School of Telemedicine and Biomedical Informatics, SGPGIMS, Lucknow.To critically evaluate currently available IP based videoconference software solutions. To design cost-effective low end telemedicine platforms integrating IP based video- conference software solution. To carry out a Proof of Concept study. All the available IP based video-conference solutions integrated into a low cost telemedicine platform using both wired and wireless broad band were included for the study. The following features were taken into account for evaluation - Quality of video and audio, Option of VoIP (Voice over Internet Protocol) and support most of the Industry standard protocols., Installation setup, user friendliness and machine dependency, conference recording, Features of desktop and application sharing, File transfer and power point sharing, Optimum bandwidth requirement. Following IP based video-conference software products were included in the study - Team viewer, Skype/iChat, Netmeeting, Webex, Polycom PVX, Vennfer and People-Link. For proof of concept study remote telemedicine partners were connected with broadband. A low cost telemedicine platform was designed based on Atom processor. Both the native camera and webcam were used for video-conference. Result of evaluation of various features of the IP based video-conference software will be discussed.Telemedicine platform designed on software based video conferencing system is ideal for rural and mobile healthcare setting and would be cost-effective over and above hardware based video conferencing solutions.

TELEMEDICINE SECURE DATA HYBRID NETWORK INTRUSION DETECTION MODEL B. Naga Malleswara Rao, N.Sambasiva Rao, V. Khanaa Bharath University, Chennai- 600 073, India [email protected], [email protected], [email protected] ABSTRACT: This paper proposes a new intrusion detection methodology based on security data. Public computer networks, such as the emerging ISDN (Integrated Services Digital Network) technology, are vulnerable to eavesdropping. There fore it is important for telemedicine applications to employ end-to-end encryption mechanisms securing the data channel from unauthorized access or modification. If the protection of data is critical, then the ability to detect and block potential intruders is an essential component of any security solutions. Over the last few years, various companies have developed software to automatically detect intrusion onto networks. These detection systems are particularly for anomaly detection or misuse of detection separately. In this paper, a hybrid intrusion detection model is proposed for intrusion detection, which can detect problems involved in pattern representation, computability, and performance. KEY WORDS: Telemedicine, IDML (Intrusion Detection Markup Language), JADE (Java Agent Development Environment) IDS (Intrusion Detection System), ISDN (Integrated Services Digital Network). 1. INTRODUCTION and network sources and analyze this information for symptoms of security breaches In telemedicine, biomedical research and clinical practice, a variety of biomedical instruments are required 2. ARCHITECTURE FLOW for the measurement and monitoring of various physical and physiological data. Security and privacy are the The model is based on the hypothesis that security growing concerns in the open distributed software violations can be detected by monitoring a system’s audit systems due to Internet’s rapid growth and the desire to records for abnormal patterns of system usage. In this conduct business over its safety. This desire has led to the work, we will collect the audit data and are mapped into advent of many security architectures and protocols, random variables using parametric mixture model. which deals with authentication, cryptography, and Mapping audit data, which is generated by the system to authorization to avoid a possible intrusion. There has random variables, is performed during extraction of been significant work in the field of intrusion detection reference data in system behaviour modelling and system that comes into picture after an attack. Most of the usage. The figure 1 shows the network, which depicts the projects are largely focused on the analysis of attacks use of IDS in the network. Groups are the users, which within a single isolated system. Incidents, however, often use the network for sending data using the network consist of a large series of widely distributed exploits, policies set by the server. The audit data and pattern data involving numerous systems, networks, operating are created for each user, which is used for comparison systems and applications. In recent years, several with the existing profiles using IDS intrusion detection systems have been designed to identify and detect possible intrusion behaviors. In 3. IMPLEMENTATION addition to intrusion prevention techniques, such as user authentication and authorization, encryption, and The system is responsible for auditing and transmitting defensive programming, intrusion detection is often used audit records to the intrusion detection system for as another wall to protect computer systems. Intrusion analysis. From the server perspective, the monitor agent Detection Markup Language (IDML) is used to develop will get the log file from the client and send the file to the rule based intrusion detection and parametric mixture comparator after converting into IDML format. IDML model algorithm is used to detect the intrusions Parser is used to convert the log file into XML format. depending on statistical based algorithms. But, currently IDML parser also helps in comparing the coming file no detection system exists which can detect both types of with the existing profiles stored in the server. Moreover, attacks simultaneously. Gaussian parametrical model may the server handles the requests of the user. The not be able to model complex data. Firewalls, comparator will compare the profiles and if any anomaly authentication, encryption and vulnerability checking can or violation of rules happened, then it will set the severity all offer improved security. But even when the computer level and send the same to the auctioned agent. Anomaly system is equipped with various technologies of firewalls events can include actual attacks against a computer and authentication procedures, it is still susceptible to system or more subtle and hence difficult to detect, attacks from hackers who take advantages of system probes that are aimed at information reconnaissance. flaws and social engineering. An intrusion detection Another challenge in ID is that of false positives. If system can collect information from variety of system

anyone enters false positive, IDS reports that an intrusion existing XML parser. We have designed a corresponding occurs. intrusion detection model based on IDML. In this model, the intrusion pattern described in IDML can be translated Group1 into a finite state machine because the structure of XML is regular expression. Furthermore, IDML documents can U1 be easily reused, and IDML can be extended to describe new intrusion pattern due to the standardized property of U2 Patt XML ern . Dat 4. CONCLUSION . a U3 The main advantage of this work is the provision of network intrusion detection to improve the security of Group2 I data transaction over the network. The profile is D converted into IDML tags to compare with the existing Netw S profiles. Design and development of the Intrusion Detection System are considered in 3 main stages namely orking normal behavior construction, anomaly detection and model update. Syste REFERENCES U1 m [1]. Y. L. Cheng and C. S. Laih, “The design and U2 Appli implementation of a distributed network intrusion . detection system with the reconnaissance ability,” . cation Audit Master’s thesis, Department of Electrical Engineering, National Cheng Kung University, Taiwan, June 2000. U3 Data [2]. Joliffe, I.T., Principal Component Analysis - 2nd Group3 Edition, Springer Verlag, 2002. Policy [3] Debar, H, Dacier, M, and Wespi, A, A Revised U1 Taxonomy for Intrusion-Detection Systems, IBM Research Report, 1999. U2 . . [4]. Cheeseman, P., and Stutz, J., “Bayesian classification (AutoClass): theory and results,” in U3 Advances in Knowledge Discovery and Data Mining, edited by U. M.Fayyad et al., California: The AAAI Figure 1 Information System Entities and IDS Press, 1996, pp. 61-83. Misuse detected [5] Stefan Axeisson. “The Base-Rate fallacy and the Difficulty of Intrusion Detectin” ACM Transaction on Asset New Rule Result information and System Security. Vol.3.3.pp.186-205. Set August 2000. Modify existing [6].Stefan Axelsson. The base-rate fallacy and its Rules Admini implications for the difficulty of intrusion detection. strator Proceedings of the 6th ACM Conference on Computer Even Anomaly and communications Security, November 1999. t Records [7] Seveb. A. Hofneyr. “A immunological Model of Gene Distributed Detection and its application to computer Security. “Ph.D thesis, Department of computer Sciences, rator Anomaly University of New Mexico, Albuerque. NM.April 1999. Activ Detected [8] Anup K.Ghosh and Aaron Schwartzbard. A study in using neural networks for anomaly and misues ity detection. In Proceedings of USENIX Security Profi Sysmposium 1999, 1999 Figure 2leArchitecture Layout It has been argued that it is actually this false alarm rate that is the limiting factor in an Intrusion Detection Systems performance [6]. The main advantage of anomaly detection system is that they can detect previously unknown attacks. Therefore, an XML based Intrusion Detection Markup Language (IDML), which can be used to express expert knowledge about intrusion patterns, and a corresponding model of an intrusion detection mechanism based on IDML is proposed here. Since XML is a standard language that is clearly understandable, so IDML is used. Thus, the IDML parser can be easily implemented by simply modifying the

[9]. Wenke Lee and Dong Xiam. Information-theoretic measures for anomaly detection. In The 2001 IEEE Symposium on Security and Privacy, Oakland, CA, May 2001.

Technical Session - 7 Emerging Diseases -2 Mapping of Cutaneous Leishmaniosis in Kerman City from 2002 - 2006 and ITS Environmental Risk Factors; by Geographical Information System Mirzazadeh A , Hajarizadeh B, Mesgarpour B, Golozar A and Holakouie Naieni K.……………..….…150 Mapping Childhood Obesity and Overweight in Greece with the use of GIS Technology Christos Chalkias, Kostas Tambalis, Glyceria Psarra, Demosthenes Panagiotakos Antigoni Faka and Labros Sidossis…………………………………………………………………….….151 GIS Study on Infectious Diseases in Cities along Asian Highway Route 9 (Mukdahan-Savannakhet-Hue) Kosum Chansiri and Sutatip Chavanavesskul……………………………………………………….……156 A Study of Morbidity Pattern and Mental Status of Geriatric Population in Lucknow’s Urban Slums Arjit Kumar, P. Bhardwaj, P. Gupta, J. P. Srivastava and K. P. Mathur……………………..………….160 The Impact of Homelessness over Physical and Mental Health of the Street Children in Dhaka Sonya Afrin, Kazi Murshida Morshed and Sarah Bashneen Suchana…………………………………..164 Spatial Statistics Analysis of JE Occurrence and Identification of Disease Hot Spots - Case Studies in a JE Endemic District of North East India Bijoy K. Handique, Kasturi Chakraborty, Jonali Goswami and Kamini K. Sarma………………………165

Mapping of Cutaneous Leishmaniosis in Kerman City from 2002 - 2006 and ITS Environmental Risk Factors; by Geographical Information System Abstract Mirzazadeh A, Hajarizadeh B, Recent reports indicated an increase in cutaneous Leishmaniosis (CL) cases. We Mesgarpour B, Golozar A and designed the study in the context of community assessment process to identify and address the major public health related issues by explore the risk map of CL Holakouie Naieni K and assessing the environmental risk factors in Kerman. All the registered CL in [email protected] the only referral center for CL from 2002 to 2006, localized on Kerman digital map. The level of data dissemination was townships. Based on data from the national statistics organization, we determined the population and calculated the incidence of CL of each township. Secondly, the highest endemic townships were observed deeply with a specific checklist to determine the environmental risk factors. 771 cases were included. All the high endemic areas were located in the east part of Kerman. The eastern township, Sarasiyab, with 123 (15.9%) cases was the most infected region. The highest endemic townships were Sarasiyab, Emam and Sarbaz with 54.9, 52.8 and 51.2 cases per 10,000, respectively. Some minor endemic areas such as Shahab, Abouzar and Shahzadeh Mohammad (South and central regions) were going to be disappeared while Shariati, Naseriyeh-Seyedi (North and North-East regions) were the new high-risk townships (P<0.01). The east and central part of Kerman, were always the high endemic regions. Some other new high-risk areas were also detected. The most environmental factors were the bare lands between the houses, ground passages and the timeworn architecture on the buildings.

MAPPING CHILDHOOD OBESITY AND OVERWEIGHT IN GREECE WITH THE USE OF GIS TECHNOLOGY Christos Chalkias1, Kostas Tambalis2,  Glyceria Psarra2, Demosthenes Panagiotakos2, Antigoni Faka1,   Labros Sidossis2 Harokopio University of Athens: Dept of Geography1, Dept of Nutrition and Dietetics2 ABSTRACT: This paper aims at presenting a methodological approach designed to integrate health statistical and spatial data, in order to highlight differences that may exist in childhood obesity and overweight rates in various areas of Greece. Thus, a spatial database based on Geographic Information Systems technology has been constructed for mapping and highlighting hot spots of childhood obesity/overweight across Greece. The geo-coding of each sample unit (primary school) was based on the postcode of the school. Population data derived from a school-based health survey carried out in 2008 in 4.165 schools of Primary Education (~85% of all Greek primary schools). Height and weight measurements from 74.928 children aged 8 to 9 years (males: 49.4%) were analysed. The gender and age specific Body Mass Index cut-off points by the International Obesity Task Force were used in order to define underweight, normal weight, overweight and obesity. The proposed methodology yields a cartographically rendered “zonation” of Greece based on percentages of obese and overweight children. The explorative analysis of childhood obesity rates shows significant hot spots in many areas of Greece as well as in some prefectures with special characteristics. GIS technology proved an efficient tool for the visualization and explorative analysis of childhood obesity across Greece. This is the first attempt to create a childhood obesity map of Greece. KEY WORDS: childhood obesity, overweight, GIS, obesity rates in Greece 1. INTRODUCTION this purpose, geospatial research for the “obesity/overweight epidemic” was implemented with The world-wide prevalence of obesity has reached the use of Geographic Information Systems technology. alarming levels. Predictive models suggest that the ratio obese-to-thin children will continue to rise in the future Geographic Information Systems (GIS) is an efficient (Kosti and Panagiotakos, 2006). If no action is taken to technology for storage, analysis and representation of counteract the trend, the number of overweight children spatiotemporal data (Chalkias, 2007). Additionally, in the European Union is expected to rise by 1.3 million spatial visualizations (and especially thematic maps) per year, with more than 300.000 of them becoming created with the use of GIS technology can be a useful obese each year (Wang and Lobstein, 2006). According tool to help local authorities and decision makers to the aforementioned trends, childhood obesity has conceptualize health problems and their spatial been recognized as an epidemic in most developed and distribution across broad geographical areas. The role of developing countries (James, 2008, Martorell et al., GIS in various health research aspects is pointed out 2000). from many researchers (among others Croner et al., 1996, Rushton, 2003, Bhowmick et al., 2007). To the best of our knowledge, only few studies (Krassas et al., 2001, Georgiadis and Nassis, 2007) have Geospatial research in health science is divided into the estimated the prevalence of childhood obesity in Greece; following general categories: most of them have drawn conclusions from selected geographic areas (Krassas et al., 2001, Tokmakidis et a) Thematic mapping of the spatial distribution of al., 2006). Moreover, even in the international literature, diseases (among others Rushton and Lolonis, 1996, the geospatial study of obesity/overweight is limited Langford et al., 1999), (Pouliou and Elliott, 2009, Rosenberger et al., 2005). b) Spatial pattern detection methods as Openshaw’s Therefore, the aim of the present study was to examine Geographical Analysis Machine (GAM - Openshaw, the prevalence of overweight and obesity, and to assess 1995), Kulldorff’s Spatial Scan Statistic (Kulldorff, the spatial distribution of childhood obesity/overweight, 1997, Kulldorff, 2001), geographically weighted in almost all 8 to 9- year-old children, in Greece. For regression (GWR Brunsdon et al., 1998, Fotheringham et al., 2002) and Exploratory Spatial Data Analysis (ESDA, Anselin, 1995, Jacquez et al., 2005).

Accordingly, in this paper we focus on the use of GIS school in the spatial background of the study area (figure technology for the thematic mapping of childhood 1). obesity across Greece. Code / Prefecture Code / Prefecture 2. METHODS AND PROCEDURES A1 Athina 43 Magnisia 2.1 Participants A2 East Attika 44 Trikala Population data derived from a national school-based health survey. Specifically, anthropometric data and A3 West Attika 51 Grevena information on age, gender, city and area were collected, between May 1 and June 15, for 2008, in A4 Pireus 52 Drama almost all schools of Primary Education (roughly 85%); schools that did not participate were from borderland 01 Etoloakarnania 53 Imathia areas, with small numbers of children. Thus, in this paper we present data from the survey of 2008 03 Viotia 54 Thessaloniki concerning 74928 students (49,4% of them are males) from more than 85% of the total primary schools of 04 Evia 55 Kavala Greece. 05 Evritania 56 Kastoria 2.2 Measurements 06 Fthiotida 57 Kilkis After the measure of height and weight (using a standardized procedure), body mass index (BMI) was 07 Fokida 58 Kozani calculated as the ratio of body weight to the square of height (kg/m2). BMI cut-off points were used by age and 11 Argolida 59 Pella sex category (according to IOTF) for underweight, normal weight (Cole et al., 2007), overweight, and obese 12 Arkadia 61 Pieria (Cole et al., 2000), as the most proper for epidemiologic studies (World Health Organization, 1995). 13 Achaia 62 Serres 2.3 GIS construction – Geocoding of primary 14 Ilia 63 Florina schools 15 Korinthia 64 Chalkidiki A GIS provides various useful features as the ability to generate new information by integrating various datasets 16 Lakonia 71 Evros sharing a compatible reference system and the ability to present these data in various thematic maps (Goodchild, 17 Messinia 72 Xanthi 1993). Thus, in order to visualize and analyse spatial and temporal patterns of childhood obesity/overweight) 21 Zakinthos 73 Rodopi the GIS technology was adopted and the following general procedures were implemented: 22 Kerkira 81 Dodecanessa a) GIS - Spatial Database creation. 23 Kefallinia 82 Cyclades b) Geocoding of primary schools according to their address 24 Lefkada 83 Lesvos c) Creation of various thematic maps of childhood obesity 31 Arta 84 Samos The spatial database was created in a GIS context with the use of ArcGIS GIS software (Arctur and Zeiler, 32 Thesprotia 85 Chios 2004). The Hellenic Geodetic Reference System (HGRS 87) was adopted as the uniform reference system for the 33 Ioannina 91 Heraklio GIS layers. This geodatabase consists of various layers representing the administrative districts of Greece and of 34 Preveza 92 Lasithi background – environmental and socioeconomic information. Furthermore, weight condition data from 41 Karditsa 93 Rethimno the survey were added in the GIS database in order to perform the geocoding of these data. For this task the 42 Larisa 94 Chania address of each school as well as the thematic map of postcode polygons of Greece were used. With this Figure 1 Study area – Greek prefectures. procedure we ensured the allocation of each primary The next phase of the presented work is dedicated to the creation of various thematic maps of childhood body weight condition in local and national scale. Next we

present thematic maps for the body weight condition of children in national level. 3. RESULTS A first reading of the thematic maps of Greece as a whole for the childhood obesity/overweight shows up some extensive areas of high rates, located primarily in the Aegean Sea and North Greece. Characteristic zones of this type are the Crete zone (Heraklio, Chania, Rethymnon, and Lasithi prefectures) the Dodecanesse prefecture and the zone of Northern Greece (prefectures of Macedonia District and Evros in the Eastern part of Thraki). The remainder of the Greece is characterized by moderate to low percentages, with a few local enclaves of high rates. These are the areas of Preveza, Messinia, Lakonia and Argolida in the mainland and the prefectures of Kefallinia and Lefkada in the Ionian Sea. Thematic maps of Greece with the different figures for childhood obesity/ overweight for all the prefectures of the Greece are represented below (figure 2). It must be noticed that the categorization of these thematic maps is based on natural brakes classification. Additionally, two classes below the national average and two classes below this value were used. Apart from this approach in national scale, a detailed analysis based on postcode districts was implemented. From this analysis detailed thematic maps of overweight/obesity rates have been constructed. Figure 3 presents the detailed obesity/overweight map of Achaia prefecture. From these maps we can locate hot-spot pockets of high rates within each prefecture. Even in prefectures with more or less low mean values these pockets of high rates are common. For example, in figure 3b (obesity map of Achaia) although the mean percentage for Achaia is below the national mean, high values is the dominant pattern in all the broader area of Patra which is the main urban centre of the prefecture. 4. DISCUSSION This study presents the spatial patterns of childhood obesity/overweight in Greece with the use of GIS technology. The first results confirm the significant percentages of childhood obesity/overweight rates across Greece. Only 11,1% of the prefectures fall below the national mean value of overweight children while the remaining 88,9% are above this value. For obesity, prefectures that are below the national mean are 20,4%, while 79,6% of the prefectures are above this. While the differences between overweight and obesity are limited, some prefectures score relatively higher rates of obesity than overweight in comparison with the national average. Figure 2 Percentages of Overweight (A), Obese (B), and A + B (C) children in Greece. (2008 survey)

As it can be seen from the thematic maps, prefectures Figure 3.: Percentages of Obese, Overweight and Sum of that score high overweight values are concentrated in the Obese and Overweight Childs in Greece. (2008 survey) North Greece as well as in coastal and insular regions. For the obesity, the highest rates are concentrated in the REFERENCES islands and in the North prefectures of the country. Islands also score high values for the Anselin, L., 1995, Local Indicators of Spatial obesity+overweight percentage. Association – LISA, Geographical Analysis, 27, 93- The major urban centres of Greece (Athina and 115. Thessaloniki) score low rates of obesity while low rates of overweight appears only in Athina and some isolated Arctur, D., and Zeiler, M., 2004, Designing areas (Xanthi, Thesprotia). Additionally, the obesity + Geodatabases: Case Studies in GIS Data Modeling, overweight % is lower than the national mean only in 6 ESRI Press. from 54 prefectures (Attika: lowest value, Etoloakarnania, Evritania, Thesprotia, Xanthi, Pieria). Bhowmick, T., Griffin, A.L., MacEachen, A.M., Variations in the spatial distribution of overweight and Kluhsman, B.C., and Lengerich, E.L., 2007, obesity could be related – among other reasons – with Informing geospatial toolset design: Understanding variations in socioeconomic conditions, educational level, environmental conditions, cultural norms and every-day practices (Tremblay et. al., 2005, Pouliou & Elliott, 2009). Generally, when comparing the current rates of obesity to the national average value the majority of the prefectures score higher. This becomes even more alarming when one sees that in the last 10 years the national mean value has been raised ~ 5%. The examination of the detailed obesity/overweight spatial patterns for the Achaia prefecture (figure 3) shows that urban and suburban areas near to the city of Patra perform high rates. In some cases (e.g. the percentage of obesity) this finding is in contrast with the overall score of the prefecture which is lower than the total average. High rates of overweight percentages are dominant in all the areal units of Achaia prefecture with the exception of the southern post code areas of the region and at the West coastal area of Patra. Furthermore, the future analysis in local level with the use of local indicators of spatial autocorrelation within GIS context could be used to identify hot spot clusters of low and high values. 5. CONCLUSIONS The mapping of childhood obesity/overweight shows significant regional differences across Greece. Moreover, the definition of spatial patterns of low and high values with the use of GIS technology provides useful information about Geographical variation of obesity/overweight in both national and local level. The findings show remarkable high values in the islands of Aegean Sea, Crete, in parts of Ionian Sea and parts of Northern Greece. Additionally, Athina prefecture scores low rates in overweight and obesity percentages. In the future, childhood obesity data might be easily analysed within GIS context in cooperation with other geographical information in order to distinguish important correlations with environmental and socioeconomic data. Moreover, the exploratory geospatial research of obesity in local level could contribute to identify local spatial clusters of obesity and overweight in Greece.

the process of cancer data exploration and analysis. Martorell, R., Kettel, K.L., Hughes, M.L., and Health and Place, 14, 576-607. Grummer-Strawn L.M., 2000, Overweight and Brunsdon, C., Fotheringham, S., and Charlton, M., 1998, obesity in preschool children from developing Geographically weighted regression. Journal of the countries. Int J Obes Relat Metab Disord, 24, 959- Royal Statistical Society: Series D (The Statistician), 967. 47, 431–443. Chalkias, C.N., 2006, Terms and concepts in Openshaw, S., 1995, Developing automated and smart geographical information science (GIS) (in Greek), spatial pattern exploration tools for geographical (Athens: ION Editions). information systems applications. Statistician, 44, 3– Cole, T., Bellizzi, M., Flegal, K., and Dietz, W., 2000, 16. Establishing a standard definition for child overweight and obesity worldwide: international Pouliou, T., Elliott, S.J., 2009, An exploratory spatial survey. BMJ, 320, 1240-1243. analysis of overweight and obesity in Canada, Cole, T., Flegal, K., Nicholls, D., and Jackson, A., 2007, Preventive Medicine 48 (4), pp. 362-367. Body mass index cut offs to define thinness in children and adolescents: international survey. BMJ, Rosenberger, R.S., Sneh, Y., Phipps, T.T., Gurvitch, R., 335, 194-198. 2005, A spatial analysis of linkages between health Croner, C., Sperling, J., and Broome, F.R., 1996, care expenditures, physical inactivity, obesity and Geographic information systems (GIS): new recreation supply, Journal of Leisure Research 37 perspectives in understanding human health and (2), pp. 216-235 environmental relationships. Statistics in Medicine, 15, 1961–1977. Rushton, G., 2003, Public health, GIS, and spatial Fotheringham, A.S., Brunsdon, C., and Charlton, M., analytic tools. Annual Review of Public Health, 24, 2002, Geographically Weighted Regression: the 43–56. analysis of spatially varying relationships, (Chichester: John Wiley and Sons). Rushton, G., and Lolonis, P., 1996, Exploratory spatial Georgiadis, G., and Nassis, G., 2007, Prevalence of analysis of birth defect rates in an urban population. overweight and obesity in a national representative Statistics in Medicine, 7, 717–726. sample of Greek children and adolescents. Eur J Clin Nutr, 61, 1072-1074. Tremblay, M., Perez, C.E., Ardem, C.I., Bryan, S.N., Goodchild, M.F., 1993, The state of GIS for Katzmarzyk, P.T., 2005. Obesity,o verweight and environmental problem solving. In Environmental ethnicity. Health Reports 16 (4), 23–34. Modelling with GIS, edited by M.F. Goodchil, B.O. Parks, and L.T. Steyaert (New York: Oxford Tokmakidis, S.P., Kasambalis, A., and Christodoulos, University Press). A.D., 2006, Fitness level of Greek primary Jacquez, G.M., Greiling, D.A., Kaufmann, A.M., 2005. schoolchildren in relationship to overweight and Design and implementation of a space–time obesity. Eur J Pediatr, 165, 867-74. intelligence system for disease surveillance. Journal of Geographical Systems, 7, 7–23. Wang, Y., and Lobstein, T., 2006, Worldwide trends in James, W.P.T., 2008, The epidemiology of obesity: the childhood overweight and obesity. Int J Pediatr size of the problem. J Intern Med, 263, 336-352. Obes, 1, 11-25. Kosti, R.I., and Panagiotakos, D.B., 2006, The epidemic of obesity in children and adolescents in the world. World Health Organization (WHO), 1995, Physical Cent Eur J Public Health, 14, 151-9. Status: the use and interpretation of anthropometry: Krassas, G.E., Tzotzas, T., Tsametis, C., and Tech. Rep, Series 854, (Geneva: WHO). Konstantinidis, T., 2001, Prevalence and trends in overweight and obesity among children and adolescents in Thessaloniki, Greece. J Pediatr Endocrinol Metab, 14, 1319-26. Kulldorff, M., 2001, Prospective time periodic geographical disease surveillance using a scan statistic. J R Stat Soc Ser A, 164, 61-72. Kulldorff, M., 1997. A spatial scan statistic. Communications in Statistics—Theory and Methods 26, 1481–1496. Langford, I.H., Leyland, A.H., Rasbash, J., and Goldstein, H., 1999, Multilevel modeling of the geographical distributions of diseases. Journal of Royal Statistical Society Series C, 48, 253–268.

GIS STUDY ON INFECTIOUS DISEASES IN CITIES ALONG ASIAN HIGHWAY ROUTE 9 (MUKDAHAN-SAVANNAKHET-HUE) Kosum Chansiri1 and Sutatip Chavanavesskul2 1 Department of Biochemistry Faculty of Medicine Srinakharinwirot University, [email protected]. 2 Department of Geography Faculty of Social Sciences Srinakharinwirot University, [email protected]. ABSTRACT: The East–West Corridor (EWC) Project is part of the wider East–West Economic Corridor linking Da Nang in Viet Nam and Mawlamyaing in Myanmar—covering Lao People’s Democratic Republic (Lao PDR), Myanmar, Thailand, and Vietnam. As a flagship project of the Greater Mekong Sub-region (GMS) Program, it was designed to improve national road number 9 linking landlocked areas in northeast Thailand to the Vietnam coast via Lao PDR. Thailand should assist neighbour countries in terms of public health prevention particularly in infectious diseases. Because of less study on infectious diseases in Thailand neighbour countries, Thailand has to give them a hand by study on infectious diseases in these areas. And Mukdahan province is set itself as an Indochina gateway because of an advantage from international transportation linkage. The second Thai-Laos friends bridge is built according to the East-West economic corridors project. Thus, Mukdahan has been found 5 kinds of infectious diseases between 2004-2008; Diarrhea, Bacillary dysentery, Food Poisoning, Dengue Fever, and Fever of unknown origin These diseases has been discovered in urban area, particularly in Mueang district and Wanyai district. Thus, it can be discussed that a problem of infectious diseases in the area will be released if public sectors and related organizations, who respond to public health policy, take this problem into an account. Moreover, they should set a mitigation policy in order to control infectious diseases problem, which may cause by free trade policy with neighbour countries. KEY WORDS: GIS, Epidemiology. ignored from countries in the GMS, the infectious diseases will be spread through the region by economic 1. INTRODUCTION and transport development. 1.1 General Instructions This study is focused on infectious diseases in the region by using GIS technique. Then, planning of infectious The East–West Corridor (EWC) Project is part of the diseases prevention will be discussed. The study area in wider East–West Economic Corridor linking Da Nang in this research covers 3 countries; Mukdahan province in Viet Nam and Mawlamyaing in Myanmar—covering Lao Thailand, Savannakhet district in Lao PDR., and Hue in People’s Democratic Republic (Lao PDR), Myanmar, Veitnam. This study area was developed from Thailand, and Vietnam. As a flagship project of the “Geographical Information System of Important Greater Mekong Sub-region (GMS) Program, it was Infections Disease in Savannakhet, Lao PDR.”, which designed to improve national road number 9 linking conducted by Institute of Asia Pacific, Srinakharinwirot landlocked areas in northeast Thailand to the Vietnam university in 2004. coast via Lao PDR. It is the second cross-border road project in the program and is in line with ADB's thrust for 1.2 Objectives regional cooperation in the transport sector. (Fig. 1) 1. To study infectious diseases along Asian highway oute According to public health development for support and no.9 in Thailand, Lao PDR. and Vietnam. improvement quality of life of people in the country, it 2. To create database on infectious diseases along will be one of the factor of country development. It can highway route no.9 in Thailand, Lao PDR., and Vietnam. be said that Thailand is a center of GMS in terms of 3. To study by using integrated research between public education, economic, and public health. Moreover, it can health aspect and GIS methodology. be claimed than surrounding countries. Thus, Thailand should assist neighbour countries in terms of public health prevention particularly in infectious diseases. Because of less study on infectious diseases in Thailand neighbour countries, Thailand has to give them a hand by study on infectious diseases in these areas. If this study is

Author’s Guidelines – HealthGIS 2009 Figure 1 Greater Mekong Sub-region; East-West Figure 3 Area Study : Mukdahan Province Economic Corridor 1.4 Frame of conceptual 1.3 Limitation This research is studied infectious diseases in studied areas; Mukdahan province in Thailand, Savannakhet district in Lao PDR., and Hue in Vietnam. (Fig.2) Because of resource limitation, this research is divided into 2 phases; Thailand phase, and Lao PDR. and Vietnam phase. Thus, this paper is being discussed on the former only (as can be shown from Fig.3 below) Figure 4.Frame of conceptual 1.5 Research Outputs Figure 2.Area Study of Research 1. Up-to-date public health database, which will be created by GIS, in Mukdahan (Thailand), Savannakhet (Lao PDR.), and Hue (Vietnam). 2. Database will be applied for studying on infectious diseases and its planning (e.g. efficiency of herb in the area for treatment) 3. Database will be applied for planning and public health management network in order to infectious diseases control. 4. Database will be benefitted for further study in relevant topic to related organization such as ministry of public health or public health studies in the institutions.

Author’s Guidelines – HealthGIS 2009 1.6 Review Literatures 2.1.1 Maps 1.6.1 Epidemiology Map of studied area (Asian highway route no.9 from Thailand to Vietnam) is used for studying by courtesy of Epidemiology is the study of factors affecting the health GISTDA, Thailand. and illness of populations, and serves as the foundation and logic of interventions made in the interest of public 2.1.2 Documents from the organization health and preventive medicine. It is considered a cornerstone methodology of public health research, and is Public health data is gathered from public health highly regarded in evidence-based medicine for identifying risk factors for disease and determining organization in studied area. optimal treatment approaches to clinical practice. In the study of communicable and non-communicable diseases, 2.1.3 Primary data is collected form field work in the the work of epidemiologists ranges from outbreak investigation to study design, data collection and analysis area. including the development of statistical models to test hypotheses and the documentation of results for 2.2 Research tools submission to peer-reviewed journals. Epidemiologists rely on a number of other scientific disciplines, such as 2.2.1 Maps in related topic. biology (to better understand disease processes), 2.2.2 PC with ARCGIS version 9.3 program. Geographic Information Science (to store data and map disease patterns) and social science disciplines (to better 2.3 Research collecting processes understand proximate and distal risk factors). 2.3.1 Contact to related organization for related data in the studied area. 2.3.2 Online data in related topic is collected for this research. 2.3.3 Field work in the studied area is ministered for rechecking on data correction. 1.6.2 Geographic Information System (GIS) 2.4 Data Analysis The GIS is a rapidly growing technological field that 2.4.1 Data processing incorporates graphical features with tabular data in order Related data in the research area is ministered, inputted, to assess real-world problems. What is now the GIS field and rechecked by using GIS program. began around 1960, with the discovery that maps could 2.4.2 Data analysis be programmed using simple code and then stored in a Public health data is linked to physical data in the studied computer allowing for future modification when area. These data are analyzed about relation of infectious necessary. This was a welcome change from the era of diseases on the area. hand cartography when maps had to be painstakingly created by hand; even small changes could require the 3. CONCLUSIONS creation of a new map. The earliest version of a GIS was known as computer cartography and involved simple line Mukdahan province is located in north-eastern region of work to represent land features. From that evolved the Thailand. It has area 4,339.83 sq km. This area has concept of overlaying different mapped features on top of plateau landform in northern and southern part of the each other to determine patterns and causes of spatial province and Pupan mountain in the west side. The phenomenon. Khong river is laid on the eastern part of the area, which is a natural boundary between Thailand and Laos PDR. GIS technology can be used for scientific investigations, Forest area is occupied around 30% of the whole resource management, asset management, archaeology, province. The average temperature in the area is 27.9 c environmental impact assessment, urban planning, and average annual rainfall is 1,300.3 mm. cartography, criminology, geographic history, marketing, logistics, Prospectively Mapping, and other purposes. For Mukdahan province is set itself as an Indochina gateway example, GIS might allow emergency planners to easily because of an advantage from international transportation calculate emergency response times in the event of a linkage. The second Thai-Laos friendship bridge is built natural disaster, GIS might be used to find wetlands that according to the East-West economic corridors project. need protection from pollution, or GIS can be used by a The province has an advantage for economic growth not company to site a new business location to take only good transportation linkage but also it sets itself as advantage of a previously under-served market. an international distribution center. As a result of these projects, Mukdahan has to prepare for both economic 2. METHODOLOGY growth management and social change management, which infectious diseases prevention could be included in This research will be ministered as follows: the plan. 2.1 Research Sources From the research can be found that Mukdahan has been found 5 kinds of infectious diseases between 2004 and This research is studied by using various sources such as 2008; Diarrhea, Bacillary dysentery, Food Poisoning, Dengue Fever, and Fever of unknown origin. These related documents, textbooks, journals.

Author’s Guidelines – HealthGIS 2009 diseases have been discovered in urban area, particularly in Mueang district and Wanyai district. In 2006, Diarrhea has spreaded through most of area except Nongsang district and Dontarn district but it then expanded across the area in 2007. On the other hand, Dengue Fever has found in the low rate, not over 5 people per year, because these diseases have watched over from provincial public health organization. It results to low rate of the diseases. Thus, it can be discussed that a problem of infectious (C) diseases in the area will be (D) released if public sectors and related organizations, who respond to public health policy, take this problem into an account. Moreover, they should set a mitigation policy in order to control infectious diseases problem, which may cause by free trade policy with neighbour countries. (E) Figure 5 (A-E): Infectious Disease in Mukdahan during 2004-2008. ACKNOWLEDGEMENTS (A) This research would like to thank to Srinakharinwirot (B) university for research funding. REFERENCES Department of Communicable Disease Control. 2001. A strategic plan for international infectious diseases control and prevention: under health development plan in the IX national economic and social development plan (2002-2006). Ministry of Public Health, Bangkok. ____. 2004. Planning of infectious diseases prevention in border area. Ministry of Public Health, Bangkok. Lerksamran, Tawee. No date. Diseases and prevention. Ramkhamhaeng University, Bangkok. Soontornsema, Prawit and Pichainarong, Naraporn. 1988.

Author’s Guidelines – HealthGIS 2009 Epidemiology and infectious diseases control. Faculty of Public Health, Mahidol University, Bangkok. Saenawongse, Piset. 2004. Local geography along the Khong river story. Srinakharinwirot university journal of geography, 2004, pp.23-39. Geographic information system, 2009. http://en.wikipedia.org/wiki/Geographic_infor mation_system

A STUDY OF MORBIDITY PATTERN AND MENTAL STATUS OF GERIATRIC POPULATION IN LUCKNOW’S URBAN SLUMS Arjit Kumar, P. Bhardwaj, P. Gupta, J.P. Srivastava and K.P. Mathur Department of Community Medicine, Era,s Lucknow Medical College, Lucknow [email protected] ABSTRACT: To see the morbidity pattern and depression in relation to family support in Geriatric population of urban slums of Lucknow district of India having different cultural and geographical identity. The objective of the study was to see the morbidity pattern in Lucknow city and to see the depression level in relation to the family support. The study was a questionnaire based survey of the people above the age of 60 years who attended the Out Patients Department of Eras Lucknow Medical College and Hospital, Lucknow Uttar Pradesh. The survey was conducted on 210 aged people above the age of 60 years. The morbidity pattern was 67.6% having single ailment, 30.4% people having 2 ailments and only 2.0% having more than 3 ailments. 85.7% of the geriatric population was having family support and only 14.2% were without any family support. Out of the people having family support only 20.0% were depressed as against 60.0% depressed people in the category having no family support. The details are discussed in the paper. It is concluded that the joint family system and strong family ties have some positive effect on the well being of the elderly and the morbidity pattern is quite different from the other studies conducted KEY WORDS: Geriatrics, Mental health, Morbidities. 1. INTRODUCTION and depressed. In the city of nawabs, people leave their homes for the greener postures to different cities even Ageing is an irreversible process. In the words of Seneca, outside the country to seek better employment leaving “old age is an incurable disease”. More recently Sir behind the aged ones to fend for themselves.Old age is James Sterling Ross commented, “you do not heal old not a disease in itself but the elderly are vulnerable to age, you protect it, you promote it and you extend it”. long term diseases of insidious onset such as Expectation of life at birth has increased in recent years. cardiovascular diseases, cancer, diabetes, musculoskeletal The expected life projected in 2011 – 2016 has been 67 and mental disorders. There are multiple symptoms due years for male and 69 years for female. United Nation has to decline in the functioning of various body functions.. indicated that 21% of the Indian population will be above 60 years by 2050. 2. MATERIAL AND METHODS Industrialization, urbanization, education and exposure to A cross sectional study was conducted on the pre-tested western life style are bringing changes in values of life. questionnaire specially designed for the assessment of the Despite the strong family ties in India in general and morbidity pattern of the geriatric population. The survey lucknow city in particular, the old age population has was conducted in the Out Patient Department of Eras become vulnerable due to which they become distressed lucknow medical college and hospital, The purpose of the

study was explained and the confidentiality assured.The As evident from the table (Table II) the majority of the anthropometric data was collected and a detailed history aged population were having family support as 85.7% was taken regarding depression, family support, appetite, (n=180) were having family support and 14.3% (n=30) sleep, memory status, etc. Separate diagnostic criterions were not having any family support. were used for the diagnosis of different ailments. If the sleep was less than 4 hours/night it was taken as 90.00% 86% 80% With family disturbed.The data thus collected was calculated with 80.00% 7% 60% support cross tabulation on hp personal computer using Instat-3 70.00% Without family programme for statistical calculations. Mean, standard 60.00% 40% support deviation (SD), standard error of mean (SEM), maximum 50.00% 20% and minimum were calculated for age, weight, height, 40.00% body mass index (BMI), correlation between family 30.00% support and depression etc. The p-value was calculated 20.00% by applying Fisher's exact test of the significance. 10.00% 3. RESULTS 0.00% A total of 210 people above the age of 60 years were Figure 2 Family support in relation to depression recruited for the study. Out of 210 aged persons 122 were female and 88 male with mean age 70.53 (SD 7.32) of the The depression was correlated with family support as population. The mean age of female was 69.75 (SD 7.43) only 20% (n=36) of aged population having family and that of male it was 71.6 (SD 7.12). The mean body support was found depressed as against 60% (n=18) mass index (BMI) of male was 22.73 (SD 4.17) and that depressed people amongst those without family support. of female it was 25.91 (SD 4.76). Out of 210 aged people After applying the Fisher's exact test the two sided p 48 (22.8%) were from rural areas, 116 (55.2%) from value is <0.0001, considered extremely significant. The row/column association is statistically significant, which urban areas and 46 (21.9%) belonged to urban slums of gives the impression that family support leads to healthy life. During the study it was found that 56% (n=118, male Lucknow city Figure 1: Anthropometric data. 36 and female 82) people were having impaired vision and 34.2% (n=72) people were having impaired hearing. 160 Hypertension being the major problem of the aged and 120 the same was found during the current study as 79.0% (n=166, male 66 and female 100) people were having 80 raised blood pressure and majority was of the females. 40 Out of 210 aged persons studied 136 (64.7%) persons were having disturbed sleep while as 134 (63.8%) 0 Male persons were having normal appetite. Female Total Having sNUoMoreNnmbaMoeasSneiinaammnnsUngccoolRrkeokouuBbereSarmrmmaMnslDssIee Figure 1 Anthropometric data, Lucknow city 4. MORBIDITY PATTERN Family support and Depression: The morbidity pattern was studied among all persons and only 4 (2.0%) persons were having 3 or more diseases where as 64 (30.4%) population were suffering from 2 diseases. The majority of the population 67.6% (n=142) were having only single ailment (Table III). The different

ailments of the aged were also taken into consideration availability of a large workforce for occupation. In joint and it was found that majority of the population were family system there exists a strong differentiation of suffering from musculoskeletal disorders amounting to authority across generations, and a relatively passive role 44.7% (n=94), followed by 17.1% (n=36) suffering from of females. In the urban areas there is distribution of joint gastrointestinal disorders. Genitourinary ailments were family system. On of the main consequences of nuclear present 15.2% (n=32) persons with majority having family is loss of ‘elderly power' over the younger benign prostatic hyperplasia (BPH) and respiratory generation (Thornton A et al 1987). The nucleation leads disorders were present in 11.4% (n=24) persons . to a decrease in co-residence of the elderly with adult children and therefore a decrease in care and support for Percentage of distribution the aged. Mason has suggested that urbanization would lead to nucleation of family system in developing 50 44.7 Percentage of countries and a decrease in the support of elderly (Mason 40 distribution KO 1992). 30 20 17.115.211.411.410.4 The current study shows the importance of joint family 10 system as persons with family support were less prone to depression as against the persons without family support 0 and the depression was 3 times higher in the people without family support 60% against 20%. As far as the GMcauasMsritcrsRdicuoeUlieosrlonlvpostiaagrkesnseaectletniuieoolnttruaaaaryllls gender is concerned the prevalence of depression was twice more in female subjects as compared to male Figure 4 Disease wise distributions subjects (32.7% vs 15.4%) which is in accordance to the study conducted in Pakistan by Taqui et al where 33% 5. DISCUSSION female elderly were found depressed as against 15.7% male population (Taqui AM et al 2007). Also certain From the present study it is evident that the depression reviews showed the females being significant risk factors of the aged population is directly proportional to the for depression in elderly (Dfernes JK 2006, Cole MG et family support as minimum depression was found in the al 2003). people having family support as 20% of the population having family support were suffering from the depression 6. MORBIDITY PROFILE while as 60% of the population was depressed who were not having any kind of family support. After applying The present study showed 79% of elderly population was Fisher's test of significance the two sided p – value is found having hypertension while as in a study conducted <0.0001 considered to be highly significant. The in Udaipur, India the incidence of hypertension was 48% prevalence of depression in the present study was 25.7%. (Prakash R et al 2004). In a study by Hanger et al The prevalence of depression in Caucasian elderly reported Christ Church study of elderly observed population in the West varies from 1% to 42% (Djernes J prevalence of hypertension at 43.6% (Hanger et al 1990). K 2006). As far as the developing countries are Joshi et al in their study from Northern India have concerned the literature is not sufficient. The prevalence reported83% of the elderly population having more than rate of depression in a community samples of elderly in three morbidities (Joshi K et al 2003). In the present India have varied from 6% to 50% (Venkoba RA 1993, study majority of the aged 67.6% were having only single Nandi PS et al 1997). The traditional family system in morbidity which is quite different and the difference can South Asia is the joint family system (Mason KO 1992). be attributed to the vast difference in the culture and The greater proportion of the population in India in geography of the region. In a similar study on a general and in lucknow in particular live in rural areas community dwelling in Israel, Fushs et al have reported (80%) and the main occupation being farming. The joint 22.2% of the Israeli Jewish population aged 75-94 years family system is predominant in rural areas and one of having single ailment while as 20.1% had two ailments the main advantages of joint family system is the and 44.4% had three or more ailments (Fushs Z et al 1998). In our study 30.4% aged people were having two morbidities and only 2.0% people were having three or more ailments.

Kishore and Garg have reported the commonest 2. Djernes J K - 2006 Prevalence and predictors of morbidity in the elderly in the rural areas were cataract in Depression on population of elderly: A review Acta 30%, arthritis and arthralgia in 15.6%, anaemia in 13.3%, Psychiat Scand 113(5):372-387 respiratory ailment in 7.3% (Kishore S et al 1992). In the present study the majority of the population above 60 3. Fuchs Z, Blumstein T, Novikov I et al - 1998 years of age were suffering from musculoskeletal Morbidity, co-morbidity and their association with disorders(44.7%), followed by gastrointestinal ailments diability among community dwelling oldest-old in Israel J (17.1%), urogenital disorders (15.2%), respiratory Garentol Med Sci 53A:447-455 ailments (11.4%), cardiovascular disorders (11.4%), diabetes (6.6%) and the other miscellaneous ailments 4. Hanger HC, Saisbury R - 1990 Screening the elderly a were found in 10.4% of the people. These results are in Christ Church study The NZ Med J Oct 1990 accordance to the study conducted by Sunder et al in Rohtak district of Haryana (India) where 51.8% of the 5. Joshi K, Kumar R, Avasthi A - 2003 Morbidity profile elderly population was found to be suffering from joint and its relationship with disability and psychological pain and 58.0% having breathing problem (Sunder L et al distress among elderly population in North India 1999). Sunder et al have also reported visual impairment International J Epidemiology 32:978-987 in 65% of the elderly (Sunder L et al 1999) which is in accordance to the present study where 56.2% of the 6. Kishore S, Garg BS - 1992 Sociomedical problems of elderly population was having visual impairment, also the aged population in rural of Wardha district Indian J hearing impairment was found in 34.2% of the aged Public Health 41:43-48 population. Joshi et al have reported the visual impairment in 61% of the population above the age of 60 7. Mason K O - 1992 Family change and support of the years (Joshi K et al 2003). elderly in Asia: What do we know Asia Pacific Popul J 7(3):13-32 7. CONCLUSION 8. Nandi PS, Banerjee G, Mukherjee S et al - 1997 A The present study found that residing in a nuclear family study of psychiatric morbidity in an elderly population in is a strong independent predictor of depression among a rural community in West Bengal Indian Journal of elderly. The morbidity profile in our study was quite Psychiatry 39:122-129 different to what has been reported from India and other parts of the world. The degenerative diseases such as 9. Prakash R, Choudhary SK, Singh US - 2004 A study of arthritis have been found more prevalent in elderly morbidity pattern among geriatric population in a urban population while as cardiovascular ailments are not so area of Udaipur Rajasthan. Indian J of Community high. A study on a large sample size is needed in this part Medicine 29(1):35-40 of the world having quite different culture and geographical location as compared to other parts of Asia. 10. Sunder L, Chadha SL, Bhatia PC - 1999 A study on senior citizens in rural areas Health for the Millions ACKNOWLEDGEMENT 25:18-20 The authors are highly thankful to Dr Farzana Mahdi who 11. Taqui AM, Itrat A et al - 2002 Depression in the helped us a lot in the field work management and my elderly \"Does family system play a role\". A cross head Dr J P srivastava who had always been a source of sectional study. BMC Psychiatry 7:57 knowledge and endurance. 12. Thornton A, Fricke TE - 1987 Social change and the REFERENCES family: Comparative perspective from the West China and South Asia. Sociological Forum 2(4):746-779 1. Cole MG, Dendukuri N - 2003 Risk factors for depression among elderly community subjects: A 13. Venkoba R A - 1993 Psychiatry of old age in India systematic review and meta analysis. Am J Psychiatry International Review of Psychiatry 5:165-170 160(6):1147-1156



The Impact of Homelessness over Physical and Mental Health of the Street Children in Dhaka Abstract Sonya Afrin Shelter is a fundamental human right. Homeless children are being denied this Kazi Murshida Morshed right. They need supportive physical environments to contribute to their optimal and Sarah Bashneen Suchana development. For poor urban children in particular, the physical environment can present major problems, undermining their well being and their prospects for the future. The availability of decent housing, the provision of water and sanitation, the quality of space for play, the levels of traffic and pollution – these features of urban life, and many others, have profound impacts on children. These children are particularly vulnerable to the elements and as a result, many suffer from respiratory disorders and skin ailments. They are also extremely vulnerable to physical and sexual abuse and exploitation. In a survey in Bangladesh, 20% of the street children interviewed complained of rape whilst sleeping on the streets. They face harassment by the police for lack of identity documents and proof of permanent address (which can also affect their access to services such as healthcare).Lack of safe storage for belongings (e.g. shoe-shine equipment) leads to extra expenditure for ‘security as the children have nowhere safe to keep any money, they need to spend it quickly, thus hampering their life choices and longer-term development. This paper tries to study the impact of homelessness on health on street children in developing countries like Bangladesh, particularly in the city of Dhaka. Common sense and common images of homelessness lead us to believe that ‘‘life on the street’’ is extremely undesirable, with both physical and psychological downsides. The Government Project entitled ‘Appropriate Resources for Improving Street Children’s Environment (ARISE)’ reports that there are 500,000 children living on the street in Bangladesh, of which 75 per cent are in Dhaka. In Bangladesh locations where the homeless people are found on the census night, in the rail station, launch ghat (terminal), bus station, hat bazaar (market), mazar (shrine), staircase of public/government buildings, open space, etc. Illegality or lack of tenure is a key feature of urban squatter settlements. Threats and fear of eviction are commonplace. Resettlement schemes rarely work, because the old land often is convenient to work opportunities in the center city, and new areas tend to be farther out on the periphery. The impact of homelessness on street children indicates that street children confront serious threats to their well-being. Of particular concern are health problems, hunger and poor nutrition, developmental delays, psychological problems, and educational underachievement. This article examines the problems faced by homeless children in each of these areas and focus on mechanisms that can be influenced by social policy, namely, inadequate shelter conditions, instability of shelters and residences, lack of adequate services, and barriers to accessing available services.

SPATIAL STATISTICS ANALYSIS OF JE OCCURRENCE AND IDENTIFICATION OF DISEASE HOT SPOTS - CASE STUDIES IN A JE ENDEMIC DISTRICT OF NORTH EAST INDIA Bijoy K. Handique*, Kasturi Chakraborty, Jonali Goswami and Kamini K. Sarma North Eastern Space Applications Centre, Dept. of Space, Govt. of India, Umiam, 793103 Meghalaya, India *[email protected], [email protected], [email protected], [email protected] ABSTRACT: An application of spatial statistics analysis to prioritise Japanese Encephalitis (JE) incidence hotspot in a JE endemic district of Assam has been demonstrated. Spatial analytical techniques have been employed to analyse the spatial order and association of JE reporting villages in the district. Data on historical morbidity pattern of JE collected at village level provided the required stratification base for delineating the JE incidence hot spots. Spatial statistics parameters such as mean centre, standard deviational ellipse and spatial autocorrelation have been calculated for JE reporting villages. Strong spatial autocorrelation (p<0.01) among the JE reporting villages have been observed in terms of morbidity pattern as indicated by Moran’s I index. General G statistics has been calculated to categorize JE prone villages and tested for statistical significance. Based on this G statistics, JE hot spots in the study district could be identified for taking timely intervention measures with optimum utilisation of man and material. KEY WORDS: Japanese Encephalitis, GIS, Spatial Autocorrelation, Moran’s I index, G statistics, Hot spot 1. INTRODUCTION be followed to prioritise these areas of importance or hot spots with sound statistical base. Status of spatial Japanese Encephalitis (JE) is a dreaded vector delimitation, forecasting and control of JE in India has (mosquito) borne viral disease mostly prevalent in Asian been discussed in detail by Sabesan et al. (2008). countries including India. Since the first record of JE case in India in 1955 in Tamil Nadu followed by isolation of 2. MATERIALS AND METHOD JE virus from wild caught mosquitoes in 1956, in the last couple of decades, epidemics of JE have occurred in the 2.1 Study Area states of West Bengal, Assam, Manipur, Nagaland, Uttar Pradesh, Bihar and Goa in addition to South India (Khan The study was carried out in Dibrugarh district of Assam et al., 1996). JE cases have attained alarming proportions state located in the north eastern part of India considering to pose as a major public health problem in India, more the severity of impact of JE and its perennial occurrence. so due to unavailability of any cure for the disease and Average annual case load in Assam during the last two due to its very high case versus fatality ratio. decades since 1980 has been 295, the average annual incidence per million population being 12.5. The district Recent advances in Geographic Information System covering a geographical area of 7023.9 sq km lies (GIS) along with satellite remote sensing have added new between 27o 15/ N - 28° N Latitude and 94° 45/ E - 96° E dimensions to spatial statistics analysis in Longitude (Figure 1). epidemiological studies. Many studies have applied these advanced tools in understanding the host-vector Primary Health Centre Blocks in Dibrugarh relationship and their spatial distribution (Barnes & Cibula, 1979, Glass et al., 1995, Hendrickxy et al., 1999, Barbaruah Lahowal Panitola Dhiman, 2000, Abelardo et al., 2000, Jeganathan et al., 2001, Nageswara Rao et al., 2004, Handique et al., Khowang Naharani Tengakhat 2005). Spatial analytical techniques help in analyzing the spatial order and association of a variable under study. In areas like ecology, epidemiology, geology and image processing, it is often not appropriate to randomize, block and replicate the data because of the spatial associations of attribute features associated with study variable. On the other hand, it is required to stratify and prioritise areas under a particular administrative unit for better planning and managing resources. Hence a sound technique has to Figure 1 Location map of the study area

The district is divided into six blocks under six Primary area and extension around which JE cases are occurring. Health Centres (PHCs) viz. Barbaruah, Lahowal, Mean centre in terms of geographic latitude and longitude Panitola, Tengakhat, Khowang and Naharani for are measured as- monitoring and providing health care services in the district.   n xi n yi   xmc , ymc  i 1 , i1  2.2 Collection of JE case data   n  n Data pertaining to the JE cases were collected from  the office of the Joint Director of Health Services in  Dibrugarh district and from the office of the Directorate of Health Services, Guwahati, Assam. Average disease (1) incidence per unit population was calculated based on the PHC wise census records collected from the respective Where, xmc , ymc are coordinates of the mean centre, xi, PHCs compiled under National Vector Borne Disease Control Programme (NVBDCP). Data on socio-economic and yi are coordinates of point i and n is the number of status viz., poultry, cattle and pig farming practice; flood points/polygons. proneness of the areas and personal protection measures adopted for prevention of mosquito bites, etc. were 2.3.2 Directional Distribution (Standard Deviational collected through a socio-economic survey. But the Ellipse) analysis of these socio-economic data is beyond the scope of the present study. The flow of information from the Directional distribution (Standard Deviational El- periphery to the JE incidence recording units mentioned lipse) measures whether a distribution of features exhibits above is shown in the following flow chart (Figure 2). a directional trend (whether features are farther from a specified point in one direction than in another direction). 2.3 Spatial Statistics Analysis The standard distance circle shows the spatial spread of a set of point locations. The steps in deriving the standard deviational eclipse are as follows (Chou, 1997). Figure 2 Information flow of JE cases 1. Calculate the coordinates of the mean centre (xmc, ymc). This will be used as the origin in the Suspected transformed coordinate system. JE cases 2. For each point, pi in the distribution transform its coordinate by- Primary Health Private Health xi  xi  xmc Structure (Govt.) Structure yi  yi  ymc Patient Patient referred referre d Assam Medical College Sample RMRC, (ICMR) 3. calculate the angle of rotation  such that & Hospital Results Dibrugarh AMCH Medical  tan  n xi2  n yi2     nxi2  n yi2 2  4 n x Records Department i 1 i1 i 1 i1 i 1 Directorate of Health  nn Services, Hengerabari,  2 xi yi Guwahati, Assam i1 i1 Joint Director of Health Services, (2) Dibrugarh district Different analyses in GIS domain have been performed With  from the step 3 we can calculate the devia- using ARC GIS 9. Following spatial statistics parameters have been calculated for the JE reporting villages- tion along x and y axes in the following manner- 2.3.1 Spatial mean centre n  xi sin   yi cos  2 i 1 The Mean Centre (MC) or spatial mean centre gives y  the average location of set of points. Here, locations of x  n the villages having JE cases are considered for measuring the mean centre of their locations. This will indicate the  n xi cos  yi sin  2 i 1 n

(3) autocorrelation and the spatial pattern is considered to be random. 2.3.3 Measure of spatial autocorrelation Spatial statistics tool in Arc GIS software used to mea- It is of interest to see the spatial distribution JE re- sure spatial autocorrelation is based not only on locations porting villages in the district. In classifying spatial pat- of the village alone or on number of JE cases alone, but terns as either clustered, dispersed or random, we can fo- on both village locations and corresponding number of JE cus on how various points or polygons are arranged. We cases simultaneously. Given a set of village and associ- can measure the similarity or dissimilarity of any pair of ated JE cases, it evaluates whether the pattern expressed neighbouring points or polygons. When these similarities is clustered, dispersed or random. A ‘Z’ score is calcu- and dissimilarities are summarised for spatial pattern, we lated to assess whether the observed clustering / disper- have the spatial autocorrelation (Lee and Wong, 2001). sion is statistically significant or not. The Z score is cal- Here, high autocorrelation would imply the occurrence of culated as- villages with higher number JE cases and the correlation is attributable to the geographic ordering of the villages. Z  O I  E I  (7) The most commonly used spatial auto-correlation statis- SD I tic, Moran’s I coefficient (Chou, 1997) is employed to measure the autocorrelation (Eq. 4-6). Moran’s I can be 2.3.4 Delineation of JE hotspots defined as- n wij (xi  x)(x j  x) Moran’s I have well-established statistical properties W (xi  x)2   I  to describe spatial autocorrelation globally. However, it is not effective in identifying different type of clustering (4) spatial patterns. These patterns are sometimes described as ‘hot spots’ and ‘cold spots’. If high values are close to each other, Moran’s I will indicate relatively high where, W = i  j wij positive spatial autocorrelation. The clusters of high values may be labeled as a hot spots. But high positive Here, Euclidean distance is used to define the weights wij. spatial autocorrelation indicated by Moran’s I could be Corresponding to each pair of sample points i and j, let dij represent the distance between them. The distance weight created by low values close to each other. This type of is applied in an inverse manner, since the intensity of spatial relationship diminishes when the distance clusters can be described as cold spot. The G statistics increases. Hence wij = 1 / dij. When no spatial autocorrelation exists, the expected (Getis and Ord, 1992) has the advantage of detecting the value of Moran’s I is presence of hot spots or cold spots over the entire study area (Eq. 8-11). A local measure of spatial autocorrelation is the local version of the General G statistics. The local G statistics is derived for each aerial unit to indicate how the values of aerial units of concern 1 is associated with the values of surrounding aerial units  1) E(I )   (n (5) defined by a distance threshold d. The Local G statistics is defined as: n 2W1  nW2  3W 2  wij  d x j W2 (n 2  1) Gi  d   j ; i j Var (I )  xj (8) (6) j     where, W1  12 i j wij  w ji 2 This G statistics is defined by a distance d, within which the aerial units can be regarded as neighbours of i. The i  i wij  j w ji  2 weight wij(d) is 1 if aerial unit j is within d, or 0   W2  otherwise. Thus the weight matrix is essentially a binary symmetrical matrix, but the neighbouring relationship is defined by distance, d. The sum of the weights is : Here, n is the total number of geographic units (villages), Wi  wij  d  for j  i xi denotes number JE cases corresponding to ith sample j point. (9) The value of the Moran’s I coefficient ranges between -1 and 1. A larger positive value implies a clustered pattern, Basically, the numerator of (8) which indicates the while a negative value significantly different from 0 is magnitude of Gi (d) statistics will be large if neighbouring associated with scattered pattern. When the I coefficient features (No of JE cases) are large and small if is not significantly different from 0, there is no spatial neighbouring values are small. A moderate level of Gi(d)

reflects spatial association of high and moderate values, and a low level of Gi(d) indicates spatial association of low and below average values. Before calculating the G statistics we need to define a distance d, within which aerial units will be regarded as neighbours. In this exercise we have defined d as a distance of 500 meter based on the extent of the study area and spatial distribution of villages. So the Village points will be regarded as neighbours if they are within an aerial distance of 500 meter. For detail interpretation of the general G statistics we have to rely on its expected value and standardised score (Z score). To derive Z score and to test for the significance of the Figure 3 Autocorrelation among of JE reporting villages general G statistics, we have to know the expected value Mean centres of JE case distribution within the Dibrugarh district has been computed for the years 1985-2005 at of Gi(d) and its variance. The expected value of Gi(d) is- five yearly intervals. The observation shows location of mean centre at the border of Lahowal and Panitola PHCs EGi    Wi 1 (10) in 1985, gradually shifting to a tri-junction of Lahowal, n Barbaruah and Khowang PHCs during 1990, 1995 and where, Wi  wij  d  2000 and shifting further south near a quadri-junction of j Tengakhat, Lahowal, Khowang and Naharani PHCs The expected value of Gi(d) indicates the value of Gi(d) if (Figure 4) there is no significant spatial association or if the level of Gi(d) is average. Then we need to derive the Z score of A visible shift in directional distribution of JE cases the observed statistics based on the variance. According has been observed during the period 1985-2005. The to Getis and Ord (1992) the variance of Gi(d) is disease is seen to shift to new areas towards the Naharani and Tengakhat PHCs during recent years. Five standard  Var Gi   E Gi2  EGi 2 deviational ellipses drawn at five yearly intervals show a (11) gradual directional shift towards the south eastern part of the district. Moreover, JE case distribution is observed to where, spread to wider areas in successive years (Figure 4). However, no valid and plausible explanation has been    E Gi 2 Wi  n 1 Wi  x 2  1  n  j  W i Wi tfho1iusnldimtoiteddesscturidbye. the phenomenon within the purview of   n  1  n  1 n  2   2 j      x 2 j   j Where, n denotes the number of aerial units (villages) in the entire study area. 3. RESULTS AND DISCUSSION Figure 4 Spatial mean centre and standard deviational ellipses of JE reporting villages 3.1 Spatial pattern of progression of JE cases 3.6 Delineation of JE hotspots and High risk areas Spatial autocorrelation among JE reporting villages have been measured with Global Moran’s I index. Global The general G statistics has been calculated to Moran’s I, O(I) calculated with all the villages reporting delineate the areas based on whether high cases tend to JE at least once during last 21 years starting from 1985 is cluster in the area. In other words it will identify the JE found to be 0.00683 with Expected value E(I) 0.00093. Z hotspots in the study district. Highest value of Gi is score is found to be 5.04, which is significant at 99% calculated to be 8.374 and the lowest is -1.908. Z Scores confidence level (p< 0.01). have been calculated for testing the statistical significance. Villages with Z score more than 2.56 has These results confirm that spatial distribution of JE been considered to significant at 99% confidence level occurrence is non-random (Figure 3) and hence calls for special attention in the clusters of high JE occurrence.

(p<0.01) and put in the hotspot category. Three REFERENCES prominent clusters have been categorised as JE hotspots. The biggest hotspot with 54 villages is observed around Abelardo, C., Moncayo, John, D. Edman and John, T. the district head quarter on the bank of Brahmaputra Finn., 2000, Application of geographic information river. The second hotspot with 4 villages has been technology in determining risk at Eastern Equine observed near Namrup tea garden. Another hotspot with Encephalomyelitis virus transmission. Journal of two villages has been observed near Ekoratoli tea garden. American Mosquito Control Association. 16(1), 28- These clusters of villages need immediate attention in 25. terms taking long term intervention measures. Barnes, C.M. and Cibula, W.G., 1979, Some implications There are 422 villages in the district which are found of remote sensing technology in insect control to have Z value more than 1.65 (significant at 90% programmes including mosquitoes. Mosquito News. confidence level). These villages are categorised as High 39, 271-282. JE risk villages (Figure 5). Chou, Y.H., 1997, Exploring Spatial Analysis in Geographic Information Systems. Onwards Press, 202-205. Dhiman, R. C., 2000, Remote sensing: a visionary tool in malaria epidemiology. ICMR bulletin. 30(11), 1-2. Figure 5 JE incidence hotspots and high JE risk villages Getis, A and Ord, J.K., 1992, The Analysis of spatial in Dibrugarh association by use of distance statistics. Geographical Analysis, 24 (3), 189-207. 4. CONCLUSION Glass, G.E., Schwartz, B.S., Morgan, J.M. III, Johnson, The study shows the potential application of spatial D.T., Noy, P. M. and Israel, E.,1995, Environmental statistics analysis to delineate the disease incidence hot risk factor for lyme disease identified with spots at village level. This will help health department Geographic Information System. American Journal of authorities to prioritise their focus on these areas for Public Health. 85, 944-948. taking timely intervention measures. It will also serve as baseline information in a particular point of time and will Handique B.K., Khan S.A., Goswami, J., Chakraborty K., help in future planning and monitoring. Detail study in Bora R., Sarma, K.K., 2005, Characterization of JE the hotspots will reveal the critical factors responsible for prone areas of Dibrugarh District of Assam using disease transmission and outbreak. Remote Sensing and GIS. Proceeding of the 25th ISRS Annual Convention and National Symposium, 6- ACKNOWLEDGEMENTS 9 Dec, 2005, Ranchi, Jharkhand, India. The authors are thankful to Dr. P. P. Nageswara Rao, Hendrickxy, G., Napala, A., Dao, B., Batawuli, D., Director, North Eastern Space Applications Centre for his DeDeken, R., Vermilien, A. and Stingenberg, J.H.W., guidance and encouragements. The authors would like to 1999, A systematic approach to area wide tsetse thank Scientists and Technical Assistants from Regional distribution and abundance maps. Bulletin of Medical Research Centre NEH (ICMR), Dibrugarh for Entomological Research. 89, 231-244. their sincere support and help during the course of study. Sincere thanks also due to Joint Director of Health Jeganathan, C., Khan, S.A., Ramesh Chandra, Singh, H., services, Govt. of Assam and Assam Medical College Srivastava, V. and Raju, P.L.N., 2001, authorities for providing detail data on JE cases in the Characterisation of malaria vector habitats using study area. remote sensing and GIS. Journal of the Indian Society of Remote Sensing. 29(2), 31-36. Khan, S.A., Narain, K., Handique, R., Dutta, P., Mahanta, J., Satyanarayana, K. and Srivastava, V.K., 1996, Role of some Environmental factors in modulating seasonal abundance of potential Japanese Encephalitis vectors in Assam. Journal of Tropical Medicine and public Health. 27 (2), 382-391. Lee, J. and Wong, D.W.S., 2001, Statistical Analysis with ARCVIEW GIS. John Wiley & Sons, 135-189.

Nageswara Rao, P.P., Elango, H.R., Rajmohan, H.R. and Krishna Murthy, V., 2004, Satellite-based assessment of physiographic disposition to health problems in Kasargod district, Kerala, India. Journal of the Indian Society of Remote Sensing, 32(1), 75-79. Sabesan S., Konuganti, H. K. R. and Perumal, V., 2008, Spatial delimitation, forecasting and control of Japanese Encephalitis: India –a case study. The Open Parasitology Journal, 2, 59-63.

Technical Session - 8 Healthcare Planning and Management-2 Spatial and Spatio-Temporal Clusters of under-Five Mortality in Rural Bangladesh M. Zahirul Haq, Nurul Alam and Peter Kim Streatfield.……………………………….………..……...…171 Spatial Analysis of Reach Settlement and Expense Transportation to the Public Health Community in Depok, West Java, Indonesia Martya Rahmaniati and Rio Perdana …………………………………….………………………………...174 Role of Socio-Cultural System in the Emergence of Diseases – a Case Study of Hyderabad M.V. Lakshmi Devi………………………………….………………………………………………….……178 Modelling Spatial Risk for Prime-Age Adult Mortality in Viet Nam Deok Ryun Kim, Mohammad Ali, Vu Dinh Thiem, Camilo J Acosta, Lorenz von Seidlein, Michael Favorov and John D Clemens ……………………………………………..…..….………………181 A Spatial Decision Support System to Blood Bank Services: A Case Study to Private Blood Bank in Tumkur Prema Sudha, Shivakumar swamy, B. S. Adiga and Smita M. Kolhar………………...………………….182 Providing the Possibility of using GIS in Management Decision Making in the Health Sector Mohammad Zare, Parvin Shamszadeh and Abbas Najjari …………………….……………..……………183 The Analysis of Accessibility Levels and Health and Therapeutic Service Allocation with use of GIS (A Case Study Zanjan City Hospitals) Mohsen Ahadnejad and Abdollah Heidari ……………………………………….………..................................184

SPATIAL AND SPATIO-TEMPORAL CLUSTERS OF UNDER-FIVE MORTALITY IN RURAL BANGLADESH M. Zahirul Haq, Nurul Alam, Peter Kim Streatfield International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR, B), Mohakhali, Dhaka-1212, Bangladesh, [email protected] , [email protected] , [email protected] ABSTRACT: These mandatory guidelines are provided for preparation of papers accepted for publication. Reproduction is made directly from author-prepared manuscripts, in electronic or hardcopy form, in A4 paper size 297mm x 210mm (11.69 x 8.27 inches). KEY WORDS: Manuscripts, Proceedings, Guidelines for Authors, Style guides, CDRO coordinates in degree decimal format using ArcGIS 1. INTRODUCTION software. The coordinates refer to the location of the cluster (village) contour. Mapping is important to identify areas that are Data on number of deaths and children aged below 5 underserved and make distribution of the services years in each village in the study period 1998-2007 were uniform. Particularly in resource poor settings obtained from the HDSS database. Registration of vital population- wide interventions are often too expensive to events (births, deaths, migration and marriage and marital implement. To maximize the benefits health planners disruption) is complete and date of the events is precise in need information to target the scarce public health HDSS area because of too frequent visits of households resources to the high-risk areas. Spatial, temporal and by female community health research workers. They space-time scan statistics are commonly used to identify visited every household monthly till 2006 and bi-monthly areas at high risk. Statistical methodology to identify the from 2007 to record vital events in standard event forms. high-risk group is under constant development. In this paper we use the Kulldorff spatial scan statistics for the 3. DATA ANALYSIS identification and testing for clusters of childhood mortality in rural Bangladesh. Knowledge of spatial Overall childhood mortality rates for each year were distribution of mortality, morbidity or service statistics is calculated and geographical distribution of the overall important to meet the needs of those that need them the child mortality was investigated using SaTScan. Spatial most. The objective of the study is to identify spatial and scan statistics (Kulldorff, 1997) was used to identify spatio-temporal villages of excess under-five mortality in clusters (villages) of under-five mortality in Matlab Matlab – a rural area of Bangladesh for the period 1998- HDSS area. SaTScan identifies a cluster at any location 2007. Estimating the mortality clustering after adjusting of any size up to maximum size, and minimizes the for the cluster-level socioeconomic indicators is the other problem of multiple statistical tests. In order to scan for objective of the paper. small to large clusters the maximum cluster size was set to 50% of the total population at risk. Scanning was set to 2. METHOD AND DATA search only for villages with high proportions of deaths. No geographic overlap was used as a default setting, so Data of the study came from the Health and Demographic secondary clusters would not overlap with the most Surveillance System (HDSS) maintained by ICDDR,B in significant clusters. To ensure sufficient statistical power, Matlab, Bangladesh. HDSS area consists of 142 villages. the number of Monte Carlo replications was set 999, and Villages are of varying size in population and area. Very clusters with statistical significance of p<0.001 were small villages (having less than 1000 population) are reported. merged with adjacent villages that are similar in village- level adult literacy and household asset scores. After 4. RESULTS merging the small villages we are left with 90 villages (or clusters) for spatial analysis. Geographic Information A total of 3,448 under-five deaths were reported in Systems (GIS) in HDSS generate spatial data; these are Matlab HDSS area for 1998-2007 with an average yearly geo-coordinates of each bari (cluster of households that rate of 11.4 deaths per 1000 under-five children (Figure share common courtyard and whose heads are related by 1). This results in cumulative rate up to age 5 for the 10- blood), tube-well, health facilities, schools, markets and year period is 56. There was a clear gradual decline in village. Spatial database (polygon) is produced for each mortality in the study period; the rate fell to 8.1 in 2007 unique cluster and picked up the real from 14.5 in 1998; 44% decline in 10 years period.

Figure 1 Death rate (and 95% CI) per 1000 under 5 5. DISCUSSION children, 1998-2007. Registration of vital events in HDSS area is complete with precise date of events. The gradual decline in under- Temporal analysis did not identify period with five mortality is real and consistent with the results of the statistically significant high mortality rate (data not nationally representative sample surveys, the Bangladesh shown), suggesting that mortality risks in villages were demographic and heath survey in 1996-2007. Spatial- distributed uniformly over the years. Spatial analysis temporal analysis of mortality before the fifth birthday reveals that two statistically significant clusters of higher identified two clusters of villages with high mortality childhood mortality rates were identified indicating non- risks with no temporal effect. Logistic regression results random distribution of childhood mortality (Figure.1 & were found consistent with SaTScan results, revealing Table 2). Each cluster comprises a set of villages; one that the clustering in mortality is real and deserves cluster is in close proximity of the big river the Meghna attention of the health planners and managers for and is habitat of fishing communities. The area is more undertaking remedial actions. exposed to the natural disaster such as river erosion and people are less education compared to the clusters with Figure 2 Cluster with <5 mortality risk, 1998-2007. average mortality rate. Another cluster (comprising set of villages) is around the bank of river “Dhalessari, and the A possible drawback of the analysis using the Kulldorff area is far away from the upazila (sub-district) town. method is that clusters are defined as circles. This feature has some implications for interpretation of the results. If a Potential risk factors available in the HDSS area are village with low mortality is surrounded by villages with religion, education, asset score and distance to the nearest high mortality, it is included in the cluster although some health centre. We considered two factors; education and characteristics of the village are different from other economic condition (measured with percent of the adults villages. If clustering of mortality is, say, along the river having education grade 5 or more and percent of a circle is not appropriate to detect it. The first limitation household belonged to the top two asset quintiles in the is similar to that of control trial cohort. The second village respectively) to examine if high mortality risks are limitation can be minimized by viewing ArcGIS map. explained by these village-level factors. To estimate their effects on village-level mortality clustering, we fitted two Identifying clusters with high mortality risk may be logistic regression models. Villages with most likely high regarded as a first step in prioritizing areas for analysis of mortality risks are coded ‘1’, villages with secondary cause and remedial actions. Causal factors may be at the high mortality risks are coded as ‘2’ and the rest are demand side (hygiene and health care seeking) and coded ‘0’. The model I contained the ‘area’ variable only supply side (availability and accessibility). In conclusion, and the model II contained village-level education and the findings indicate non-random clustering of childhood economic condition in addition to the area variable. mortality in the study area. The clusters may be considered for prioritizing the areas for follow-up public Logistic results show that the areas identified by SatScan health efforts. as high risk remained as high risks areas (Table 2). Village-level economic condition, but adult education was not associated with the mortality risk. Control for village-level adult education explained part of the high mortality risk in the areas and it was statistically significant (the deviance between the two log likelihood is distributed as Chi-square and is found significant).

ACKNOWLEDGEMENTS REFERENCE This research activity was funded by ICDDR,B and its Sankoh OA, Ye Y, Sauerborn R, Muller O, Becher H. donors which provide unrestricted support to the Centre Clustering of childhood mortality in rural for its operations and research. Current donors providing Burkina Faso. International Journal of unrestricted support include: Australian Agency for Epidemiology 2001; 30:485-492. International Development (AusAID), Government of the People’s Republic of Bangladesh, Canadian International Kullforff M. A spatial scan statistic. Communications in Development Agency (CIDA), Embassy of the Kingdom Statistics – Theory and Methods 1997; of the Netherlands (EKN), Swedish International 26:148196. Development Cooperation Agency (Sida), Swiss Agency for Development and Cooperation (SDC), and Kullforff M. Information Management Services, Inc: Department for International Development, UK (DFID). Software for the spatial and space-time We gratefully acknowledge these donors for their support and commitment to the Centre's research efforts. scan Statistics 2006. [http://www;satscan.org/]. Chaikaew N, Tripathi N K, Hara S. Exploring spatial and spatio-temporal clusters of malaria in Chiang Mai, Thailand. International Symposium on Geoinformatics for Spatial Infrastructure Development in Earth and Applied Sciences 2008. Table 1 Under-five mortality in HDSS area for 1998-2007 using spatial analysis scanning for high rates Type Location Cases Expected Relative Risk P-value Most likely 15 villages in Durgapur 723 602.3 1.25 P<0.001 Secondary 23 villages in Nayergoan 746 636.4 1.22 P<0.001 Table 2. Odds ratios (and 95% CI) of under-five mortality for areas and village-level education and economic condition Village Characteristics Model I Odds ratio (95% CI) Model II Group of villages: with average risk 1 1 with secondary high risk 1.29* (1.19-1.41) 1.23* (1.13-1.34) with most high risks 1.31* (1.20-1.42) 1.23* (1.13-1.34) Village-level characteristics: adults with education grade 5+ (%) household in top two quintiles (%) 1.001 (0.997 – 1.006) Log likelihood (df) -18873.2 (2), p<0.001 0.984 (0.979 – 0.989) -18851.4 (4), p<0.001

SPATIAL ANALYSIS OF REACH SETTLEMENT AND EXPENSE TRANS- PORTATION TO THE PUBLIC HEALTH COMMUNITY IN DEPOK CITY, WEST JAVA, IN 2008 Martya Rahmaniati, Rio Perdana Biostatistics and Demography Department, Faculty of Public Health, University of Indonesia, Indonesia Health Official of Solok City, West Sumatra, Indonesia, [email protected] ABSTRACT: One of the efforts to improve public access towards qualified health service is improving access towards basic health service. The role of public health centre, as an institution which administers health service on the first level involved with public, becomes very important. Therefore, public health centre must be close to settlement and the cost for going to public health centre must be cheap as well. This research aims at knowing the spreading pattern of public health centre by using Nearest-Neighbor Analysis Method, knowing the scope area of public health centre around settlement by using SIG model and knowing the transportation fee from the settlement to public health canter. From the result of this research, it is obtained that the spreading pattern of public health centre in Depok city is R = 1,399 (Random). From scope area to public health centre, it is obtained that 26,1 % of population are far from public health centre and 73,9 % of population are close to public health centre. In terms of fee, 26,1 % of population must pay Rp. 2000-6000 to come to public health centre and 73,9 % must pay Rp. 8000-10.000 rupiahs. KEY WORDS: Spatial, Nearest-Neighbour Analysis, Public Health Centre, Transportation. 1. INTRODUCTION 47 public health centers. However, nowadays, Depok city has only 27 public health centers. Out of 27, the 1.1 Background number of patients’ visits in 2006 is 788.062 with the average visit around 101 everyday for every public One of the efforts to improve public access towards health center. However, if we see the average distance, qualified health service is improving access towards visits per day are very various, around 18-314 visits. On basic health service. In this case, the role of public the other hand, outpatient visits to hospitals are 625.730 health center, as an institution which administers health in 2006 (Dinkes, 2007). service on the first level involved with public, becomes very important (Azwar, 1996). Public health center is 1.2 The Aim of Research responsible for administering health development in its area, that is, by improving awareness, will and Depok city, as one of the support areas of DKI Jakarta, capability of living healthy for everyone who lives in has a quite high rate of population growth. According to that area to realize the highest level of health. By doing Depok city’s center of statistics (Badan Pusat Statistik so, access towards qualified health service can be Kota Depok), the rate of population growth in 2006 was improved through the improvement of public health 3,44 %. This situation must be balanced with the ade- center performance (Depkes, 2005). quate provisions of health service facilities and the right place of those facilities, that is, close to the settlement The health service, administered by Indonesia area, will ease population to reach them. government, are public health service (Puskesmas), as a means of health service at the first level, and hospitals, Therefore, based on that fact, it is necessary to draw the with their various levels, at the second and third level. spreading of health service facilities, especially public (Azwar, 1996). health center in Depok city and the pattern of settlement scope to the public health center in Depok city. By do- Nowadays, Depok city has its own facilities/means of ing so, it will be obtained a picture of the scope area of health service, either state-owned or private-owned. public health center in terms of spatial perspective. Based on the Depok city health profile in 2007, the Based on the result of analysis on pattern of settlement numbers of public health center are 27 Puskesmas and scope, it is obtained the estimation of population’s trans- 12 hospitals. According to the working guidelines of portation fee from housing to public health center in De- public health center, the population target, which is pok city. served by a public health center is 30.000 people on average. With the number of population around 2. RESEARCH METHODOLOGY 1.420.480, ideally, Depok city should have had around