DT-Healthcare-Ebook

Breaking down AI:10 real applicationsin healthcare

Contents01The healthcare opportunity: AI and your data 302 8How does AI lead to better care? 03 11Clinical analytics in healthcare 04 20Operational analytics in healthcare 05 27Holistic analytics in healthcare 06 32Keeping up with the digital revolution

01The healthcareopportunity:AI and your data

Chapter 02 / How does AI lead to better care?Chapter 01The healthcare opportunity: AI and your dataRapid advances in digital technology are The care that patients receive—and theredefining the way businesses operate and methods in which hospitals deliver care—deliver value. Digital technologies—including has been dramatically transformed with theartificial intelligence (AI), machine learning, introduction of advanced technologies andand augmented reality, among others— the increasing availability of data. Systemsare transforming every industry. They’re of record like EMRs have laid the foundationaccelerating innovation, improving decision for intelligent healthcare. Applying advancedmaking, automating and speeding up processes, analytics to massive amounts of data fromand saving overall costs. internal and external sources like clinical analytics and environmental systems can helpThe healthcare industry is no exception. It’s health organizations glean deeper insights.being profoundly influenced by this ongoing These systems of insight are the next evolutiondigital revolution. Health leaders today of digital transformation. With better access torecognize that innovation requires moving the right data at the right time, you can reducebeyond the use of electronic medical records operating costs, improve the quality of care,(EMRs) and embracing key principles of increase the involvement of consumers in thedigital transformation. care process, and optimize provider satisfaction.Geo/Social/ Advanced Claims & cost dataEnvironmental data analytics opportunities Claims, revenue cycleWeather patterns economics,social services data Patient & citizen dataClinical data Purchasing patterns, open government data, social mediaEMRs, diagnostic images Pharma & life science data Clinical trials, genomics

Chapter 01 / The healthcare opportunity: AI and your data 5Unstructured data is the informationthat resides outside of organizeddatabases such as electronic healthrecords and lab reports. If webecome capable of tapping thepotential of that data, we couldmake patient care more efficientand cost-effective than ever before.” Source: PWC Report, 2017

Chapter 01 / The healthcare opportunity: AI and your data 6A new paradigm for betterpatient careAI has been a long-promised savior for thehealthcare industry, and many providers aregrowing skeptical. The reality is that—while thepromise of the future is very real—there arepractical applications in place now that are alreadymeeting the needs of healthcare executives.According to the Advisory Board Company’s Annual HealthCare CEO Survey, top areas of concern for hospital and healthsystem executives in 2016 were: Engaging physicians in reducing clinical variation Redesigning health system services for population health Meeting increasing consumer expectations for service Implementing patient engagement strategies Controlling avoidable utilization 53% 52% 47% 45% 44%

Chapter 01 / The healthcare opportunity: AI and your data 7The new paradigm for healthcare is the use Healthcare leaders can take advantage of AIof real-time data and artificial intelligence to tools to empower care teams, engage patients,enable a predictive and prescriptive analytical optimize clinical and operational effectiveness,approach. You might wonder what that means. and ultimately transform health.In practical terms, it means that AI is providingthe capability to sense the healthcare world,comprehend, act, and learn. AI uses machinelearning and delivers capabilities to mimichuman types of behavior and performance—toultimately improve patient care and outcomes.AI systems gather and crunch massive amountsof data in real time to identify patterns. Theythen use that information to automate andstreamline healthcare processes. Here are a fewexamples of what these systems can do:Predict the future condition of 84%individual patients based onearly warning signs. of healthcare executives believeDetermine best practices in AI will revolutionizeoperations management by the way they gaincomparing data from multiple information.facilities. Digital Health TechnologyRecommend medications Vision 2017, Accenturebased on the successfultreatment of similar patients.Detect and prevent fraudulentclaims or other abuses.

02How does AI leadto better care?

Chapter 02 / How does AI lead to better care? 9Chapter 02How does AI lead to better care?It’s important to enable AI for everyonein the care process and understand thesignificant opportunities that exist foreveryone. Transformation based on dataand AI is delivering clinical and operationalanalytics solutions that can improve thequality of care for patients throughout theircare experiences. Let’s see how:Clinical analytics focuses on Operational analytics focuses on thethe use of data and analytics use of data and analytics to improveto improve clinical treatment the efficiency or effectiveness ofprocesses and outcomes. systems used to provide and manage care processes.For example: Clinicians can pullinsights from data to help identify For example: By using AI, healthcareat-risk patients and deliver optimal teams can predict operational issuestreatments. Sophisticated analytics and track safety metrics, monitorengines enhanced through machine equipment health, maintain thelearning and AI can provide evidence integrity of the supply chain, andthat can inform actions. identify fraud.

Chapter 02 / How does AI lead to better care? 10Following are somescenarios in which AIis already deliveringvalue. They exemplifyopportunities to improveboth clinical quality andoperational effectiveness inhealthcare organizations.Clinical analytics Behavioral Population analytics healthTransform data intoprescriptive insights. Predictive care Medical image guidance intelligence Readmissions Data & AI Throughput management managementOperational analytics Cost Staffing management managementGain actionable insights tooptimize performance. Claims management

03Clinical analyticsin healthcare

Chapter 03 / Clinical analytics in healthcarePredictive care guidance Start with What is predictive these questions care guidance?Do you use analytical systems Predictive care guidance uses analyticalthat help clinicians predict the solutions to search through large amountseffectiveness of treatments? of data from sources like EMRs, smartIf so, in what areas are you applying medical devices, patient and populationpredictive care guidance? demographics, and the public domain toIf not, are there areas within the find hidden patterns and trends and predictprovision of care in which you would outcomes for individual patients. Mostfind such systems especially useful? predictive care guidance relies on AI learning models that become more precise when Key benefits additional data and cases are introduced.For providers Predictive analytics is a data-driven crystalClinical pathway prediction ball taking analytics to the next level, beyondDrug effectiveness descriptive or diagnostic methods that lookDisease progression prediction backwards on what happened and why.For payors How can it help yourHealth risk prediction organization?Predictive risk scoring Predictive care guidance enables clinicians to determine the likelihood of disease and helps with determining diagnoses and predicting future wellness or illness. Predictive guidance can improve the quality of healthcare and reduce the costs of care. It provides clinicians with answers they’re seeking for individual patients, with a focus on increasing the accuracy of diagnoses.

Chapter 03 / Clinical analytics in healthcare 13Here are some outcomes of using analyticalmodels that connect symptoms to treatments:For providers: For payors:Clinical pathway prediction. Assess and Health risk prediction. Predict thepredict which treatment option will likely likelihood that a patient presenting aproduce the best outcome for a patient. certain set of symptoms is at risk for an adverse health event.Drug effectiveness. Predict which drug willproduce the best outcome for a patient. Predictive risk scoring. Assess which patients might be at risk for readmissionsDisease progression prediction. Predict the and hospital-acquired infections.likely path and progression of a disease.Predictive care guidancein actionA new artificial intelligence tool launched by OchsnerHealth System analyzes thousands of data points topredict which patients will deteriorate in the nearfuture. Built with a machine learning platform, thetool triggers alerts to prompt Ochsner’s care teams tointervene and proactively treat patients and preventemergency situations.During a 90-day pilot with the tool, Ochsner was ableto reduce the hospital’s typical number of codes(cardiac or respiratory arrests) by 44%. In additionto sending pre-code alerts, this predictive model iscapable of predicting any patient deterioration thatneeds attention, based on lab values, vital signs, andother data.Watch the video to learn more about how Ochsnerprevents cardiac arrests by using AI.

Chapter 03 / Clinical analytics in healthcareBehavioral analytics Start with What is behavioral analytics? these questions Behavioral analytics is a term used inHow well do your care providers marketing to describe the analysis ofand care recipients conform to consumer behavior patterns that informs howrecommended care protocols? to market or deliver an action in a way thatDoes your organization increases the odds of adoption.use value-based paymentsystems, in which quality or While clinical analytics can recommend aoutcome measures affect your clinical action, applying behavioral analyticsreimbursements? increases the likelihood of the action being taken. This is sometimes called nudging the Key benefits patient or care provider.For providers Other industries use behavioral analytics toPatient engagement suggest add-on sales or display content basedReadmissions on previous usage patterns. This is how Netflix recommends movies that a customer mightFor payors like. The method of deriving suggestions isIndividual health and well-being particularly important in healthcare, because suggestions will be put off or ignored if they’re not easy to implement. How can it help your organization? Incomplete application of evidence-based medicine is a leading cause of poor outcomes and increased overall cost of care. Behavioral analytics can increase the adoption of recommended practices.

Chapter 03 / Clinical analytics in healthcare 15Here are some benefits of behavioral analytics:For providers: For payors:Patient engagement. Engage patients and Individual health and well-being. Improvecare providers to ensure conformance and health awareness and preventive care choicesincrease the likelihood of the best possible care. in patients.Readmissions. Reduce readmissions andeven prevent initial admissions throughproactive targeting. Behavioral analytics in action Behavioral analytics can drive nudge platforms, like one from Azure partner NextHealth that delivers recommendations to health plans to increase member conformance. It enables increased patient engagement and risk identification—to ultimately improve health and cost outcomes— by using an analytics platform and personalized data to provide optimized recommendations.

Chapter 03 / Clinical analytics in healthcarePopulation health Start with What is population health? these questions The term population health, which is widelyWhich of your patient populations used in the healthcare world, covers a varietyconsume the most resources? of topics.Is your organization using value-based payment systems, in which Population health strives to influence thequality or outcome measures affect delivery of care to a group of individuals thatyour reimbursements? have similar healthcare needs, as opposed to focusing on evaluating and treating medical Key benefits conditions one patient at a time.For providers How can it help yourIntegrated care organization?Specialty carePatient engagement Payment systems are moving from a fee-for- service business model to one that incorporatesFor payors value into the payment equation. MeetingCost management payor requirements during this transitionSelf-care management requires greater use of data and analytics, including better data on patient-reported outcomes, social determinants of health, patient and member risk stratification, and activity-based costing.

Chapter 03 / Clinical analytics in healthcare 17Here are some benefits Population healthof implementing in actionpopulation health: Scientists and physicians at Johns HopkinsFor providers: Medicine are gathering huge amounts of data from medical care, genomics, and wearableIntegrated care. Implement integrated devices to predict disease progression andcare—coordinated treatment across care pinpoint individual treatments. They’reteam members, including clinicians, social examining individual diseases by lookingworkers, physical therapists, and behavioral deeply into their subgroups.health care professionals. Analytics can help Because patients in the same diseaseidentify and measure the effectiveness of care subgroups are likely to have the sameacross all care settings. biological conditions and show the same response to treatments, researchers can useSpecialty care. Use data to determine the this information to discover mechanisms thatbest ways to manage health needs and drive specific diseases.outcomes for entire populations of people The team is conducting data investigations ofsuffering from chronic conditions. patients treated for prostate cancer, multiple sclerosis, cardiac arrhythmias, amyotrophicPatient engagement. Empower patients to lateral sclerosis, and more—to improvemore effectively manage their own health and diagnoses, prevention tactics, and cures.participate in the decision-making process to Watch the video to learn more about howimprove outcomes. Johns Hopkins Medicine plans to change the face of disease.For payors:Self-care management. Use patient data toimprove patients’ understanding of their rolein their wellness, help them stay healthier, andreduce the cost per service.Cost management. Manage the health ofpopulations by creating better outcomes at anefficient cost.75% Patients with chronic conditions consume more than 75 percent of healthcare spending.

Chapter 03 / Clinical analytics in healthcareMedical image intelligence Start with What is medical image these questions intelligence?Are you (or your imaging partners) Medical image intelligence is the embeddingdoing anything to embed analytics of analytical capabilities into imagesor intelligence into the medical to augment or improve diagnostic andimaging process? treatment planning processes.What type of analytics are youusing to evaluate the effectiveness How can it help yourof your imaging systems? organization? Key benefits Medical images represent one of the largest categories of unstructured dataFor providers used in healthcare.Diagnosis and treatmentoptimization Specialists like radiologists, oncologists, ophthalmologists, and others are trained toFor payors evaluate medical images to assess medicalSpeed and quality conditions, make diagnoses, and deliver treatments based on their reading of these images. Analytics technologies can increase the effectiveness of these efforts.

Chapter 03 / Clinical analytics in healthcare 19Here are some benefits of implementing medicalimage intelligence:For providers: For payors:Diagnosis and treatment optimization. Speed and quality. Improve the qualityAssist specialists in making diagnoses, improve of image analysis while driving down coststreatment planning, and increase the efficiency for payors.of these processes.Medical image intelligencein actionThe Microsoft InnerEye research project uses state-of-the-art artificial intelligence to build innovative imageanalysis technologies that help doctors treat diseases likecancer in a more targeted and effective way.InnerEye employs decision forests and deep neuralnetworks to enable medical software providers to delivertools that radiation oncologists can use in planningradiotherapy treatment. The cloud-based radiomicsservice is intended to enable the development of third-party products that better assist radiation oncologistsand dosimetrists, allowing medical experts to focus onmore detailed tasks like editing and refining results.Learn more by watching the InnerEye video overview.

04Operationalanalytics inhealthcare

Chapter 04 / Operational analytics in healthcareStaffing management Start with What is staffing management? these questions Staffing management is the process ofWhat percentage of the total modelling and predicting optimal staffingcost of running your organization levels based on factors like predicted patientinvolves staffing? volume and the type and complexity of patientsHow do you determine staffing being treated.levels for your nursing unitsor clinics? How can it help yourHow often do you face overstaffing organization?or understaffing issues? Staffing is the single largest expense of any Key benefits medical organization delivering services to patients and consumers. In most hospitals,For providers staffing represents more than half of allImproved staff retention total expenses.Staff trainingStaffing level assessment Data and AI systems allow healthcare organizations to predict staffing levels with optimal accuracy and efficiency.

Chapter 04 / Operational analytics in healthcare 22Here are some benefits Staffingof implementing staffing managementmanagement: in actionFor providers: Healthcare Employee Retention is a solution that uses client dataImproved staff retention. Improve retention of to estimate employee flight risk.hard-to-find specialty clinicians to maintain sustainable It includes a predictive model andteams. Use predictive analytics to understand when an reporting that identifies whichemployee is at risk of resigning. employees might resign based on historical information from previousStaff training. Measure the success of training team members. The solutionprograms to improve them. Good training is critical in includes technologies for datahealth care because clinician performance has a direct exploration, machine learning, andimpact on the well-being of patients. results delivery.Staffing level assessment. Predict staffing levels andskill sets to optimize them based on the number ofpatients and the level of acuity of care to be provided.

Chapter 04 / Operational analytics in healthcareClaims management Start with What is claims management? these questions Claims management is the organization, billing,What percentage of filed claims filing, updating, and processing of medicalare being denied or require rework claims related to patient diagnoses, treatments,before being paid? Are you seeing and medications.changes in this trend?What are the top reasons for claims How can it help yourbeing denied? organization? Key benefits Claims fraud, waste, and abuse are significant issues worldwide. They encompasses a wideFor providers spectrum of activities, including deceptiveClaims management billing for services not rendered, performingDenials and revenue management unnecessary medical services, and abusing payment rules by coding services at higherFor payors levels than actually performed.Claims and fraud managementDenial management Denied claims represent extraordinary administrative costs to health providers and payors. Claims analytics helps you efficiently predict patterns and detect anomalies to fight fraud and waste.

Chapter 04 / Operational analytics in healthcare 24Here are some benefits of Claims managementimplementing analytics for in actionclaims management: CGI ProperPay is a medical claim analytics solutionFor providers: that helps healthcare payers audit, detect, and prevent inaccurate payments of inpatient,Claims management. Improve the outpatient, and professional claims. It usesidentification, management, and collection of advanced algorithms to detect hidden patternspatient service revenue. and anomalies within a payer’s complete claims universe to point out which claims should be put onDenials and revenue management. Reduce hold or rejected, and which paid claims have a highthe number of denied claims, a primary possibility for recovery.contributor to bad debt that represents The solution uses perceptual intelligence, machinemillions in lost net patient revenue every year. learning, and the cloud. Learn more about CGI ProperPay by watching thisFor payors: video.Claims and fraud management. Assessthe appropriateness of a claim and predictanomalies within the entire claims datauniverse to prevent fraudulent activities. Forpayors, claims management is a critical aspectof determining whether to pay or deny claims,and of determining the rate of payment forservices a health provider bills to the payor.Denial management. Use analytics to trackthe number of claims filed by providers that apayor denies. For payors, denied claims createlarge overhead costs.

Chapter 04 / Operational analytics in healthcareCost management Start with What is cost management? these questions Cost management is a broad category relevantHow do your operating costs to all aspects of care delivery. It encompassescompare to reimbursements? evaluation of all major provider and payorWhat are your top cost drivers? systems that determine the eventual cost ofDo you have systems in place that providing and paying for health and medicalhelp you predict costs so you can services.better manage them? How can it help your Key benefits organization?For providers Most provider and payor organizationsStaffing optimization today exist on razor thin operating margins.Supply cost management In most countries, the cost of providingThroughput management and paying for care is rising faster than reimbursements increase.For payorsFraud detection Health providers and payors are using solutionsShorter patient stays that improve the management of costs whileClaims denial management maintaining the quality of services provided. Cost management solutions use analytics to evaluate and improve the efficiency of major systems used in providing health and medical services.

Chapter 04 / Operational analytics in healthcare 26Here are some benefits of implementinganalytics for cost management:For providers: For payors:Staffing optimization. Match staffing Fraud detection. Spot potentially fraudulentrequirements—typically the largest single claims, including those miscoded to generatecost—to the current and future mix of patients. higher payments.Supply cost management. Evaluate which Shorter patient stays. Predict lengths ofsupplies are most cost effective. stay for covered services to estimate and minimize costs.Throughput management. Optimize theflow of patients through facilities—typically Claims denial management. Predict whichthe second highest cost—including hospitals, claims will be denied, reducing the number ofemergency departments, operating theaters, denials and the cost of managing them.and imaging departments.

05Holistic analyticsin healthcare

Chapter 05 / Holistic analytics in healthcareReadmissions management Start with What is readmissions these questions management?What level of inpatient A hospital readmission occurs when a patientreadmissions rates is your who has been discharged from a hospital isorganization currently admitted again within a specific time interval.experiencing? Avoidable readmissions are a strong indicator of a fragmented health care system that tooHas your current performance with often leaves discharged patients confused,respect to readmissions affected consumes resources that are already stretchedyour level of reimbursement? thin, and produces higher treatment costs.Do you have a way to proactively How can it help yourassess which patients have the organization?highest risk of readmission? Analytics solutions can be used to evaluate Key benefits and predict which patients are at risk for readmission, so hospitals can create a plan toFor providers reduce that risk.Readmittance risk analysisFor payorsReduced hospital readmissions rateReduced emergency departmentbounce-back rate

Chapter 05 / Holistic analytics in healthcare 29Here are some benefits of Readmissionsimplementing readmissions management in actionmanagement: KenSci’s risk prediction platform assists healthFor providers: systems in determining population health risk, optimizing clinical results and improvingReadmittance risk analysis. “Risk rate” operational efficiency.patients before a hospital stay, as they go The platform helps answer these questions:through treatment and recovery, and as part ofthe discharge process to assess their likelihood Who is likely to get sick?of being readmitted following discharge. How sick would they get?For payors: How can care be optimized forReduced hospital readmissions rate. Use outcomes and cost?analytics to help with reporting readmissionrates and penalizing providers whose rates Sepsis accounts for considerably moreare high. hospital readmissions and related costs than any of the other medical conditions trackedReduced emergency department bounce- by the government. KenSci’s solution helpsback rate. Reduce bounce back—an inpatient executives reduce the total cost impact ofreadmission that occurs when a patient seen sepsis by identifying patterns and estimatingin an emergency department returns after patients who are at high risk. This enablesdischarge because the condition doesn’t care managers to prevent the responsibleimprove after the treatment. condition, readmission, and mortality. Learn more by watching these videos: KenSci clinical analytics on Microsoft AppSource Using machine learning and AI to reduce hospital readmissions

Chapter 05 / Holistic analytics in healthcareThroughput management Start with What is throughput these questions management?What is the average occupancy of The process known as throughput managementyour inpatient facilities? includes systems and processes used to manageHow often are your facilities the cycling of patients through a healthat capacity? organization’s physical facility.Do you ever have issues like longwait times in your emergency Scenarios include moving an inpatient throughdepartment caused by variance in the hospital or moving an urgent care or traumademand for services? case through the emergency department to either discharge or admission. Key benefits How can it help yourFor providers organization?Hospital bed optimizationImproved patient satisfaction With increasing patient demand andscores (HACPHS) constrained physical resources, optimizingShorter wait times throughput is an essential operationsImproved patient transfer process management strategy. The use of data and analytics to optimize inpatient throughput enhances patient access, reduces unit cost, and improves service levels, leading to measurable value in metrics like reduced wait times, better use of capital, and an increased return on assets. Improving throughput allows more patients to receive care at each facility, reducing the need to add expensive new facilities or beds.

Chapter 05 / Holistic analytics in healthcare 31Here are some benefits of Throughoutimplementing throughput managementmanagement: in actionFor providers: Inefficiencies leading to patient overcrowdingHospital bed optimization. Determine/predict in the emergency department can impedethe flow and progression of inpatients through a patient throughput. Nursing informaticssystem, thereby maximizing use of patient rooms and prediction can play an important roleand facilities. in improving emergency departments andImproved patient satisfaction scores staffing. Evergreen Health is using predictive(HACPHS). Correlate outcomes and experience analytics in its emergency department todata to patient journey paths. Identify align its staff with patient demand. Obtainingreengineering opportunities by observing actionable data from patient in-flows helpscorrelations and Evergreen optimize staffing, reduce patientother data. wait times, and ensure that patients receiveShorter wait times. Predict and manage the care they need when they need it.variables like staffing to improve efficiencies incare delivery in emergency departments. KenSci has also developed predictive machineImproved patient transfer process. Help staff learning models that use data analytics.spot and resolve bottlenecks in patient transfers Actionable insights from KenSci’s predictivewithin a facility (for example, a patient waiting analytics tools empower:to be admitted to intensive care, where a bed isavailable but not yet cleaned). • Nursing leadership and staffing management to facilitate appropriate staffing and maintain nurse-patient ratios. • Nurses to spend more time with patients and deliver high quality care. • Administrators to improve the emergency department environment, and to reduce both burnout and turnover. Learn more by watching this video on how predictive analytics is helping nurses manage patient throughput.

06Keeping upwith the digitalrevolution Because of bulk record keeping and strict regulatory requirements, healthcare is a data-intensive industry. AI and machine learning are helping healthcare providers and payors extract the real-time information necessary to make effective clinical and operational decisions. Improving the quality of care, identifying at-risk patients, engaging patients at all levels with personalized care, and reducing overhead costs are some of the areas in which AI and data have had the greatest impact. The digital transformation revolution is deeply embedded into the improvement of nearly every aspect of the healthcare industry, and the trajectory extends forward, limited only by the time it takes to think up, develop, and implement new models. To keep up with the latest uses of AI, visit Microsoft Health.