Predictive analytics in health insurance
9 likes5,941 views
Predictive analytics in health insurance involves extracting information from data to forecast future trends and outcomes, enabling organizations to transition from historical to forward-looking perspectives. It utilizes both real-time and batch analytics for clinical decision-making and cost control, and involves developing models based on the characteristics of policyholders to predict claims. Case studies illustrate its application in optimizing healthcare resources and improving patient outcomes by predicting disease development and hospital stays.
Predictive analytics in health insurance
- 1. PREDICTIVE ANALYTICS IN HEALTH INSURANCE Prasad Narasimhan – Technical Architect
- 2. PREDICTIVE ANALYTICS ? • Predictive analytics is the practice of extracting information from existing data sets in order to determine patterns and predict future outcomes and trends. It does not tell you what will happen in the future but forecasts what might happen in the future with an acceptable level of reliability, and includes what-if scenarios and risk assessment.
- 3. PREDICTIVE ANALYTICS & BIG DATA • Predictive analytics is an enabler of big data: Businesses collect vast amounts of real-time customer data and predictive analytics uses this historical data, combined with customer insight, to predict future events. Predictive analytics enable organizations to use big data (both stored and real-time) to move from a historical view to a forward-looking perspective of the customer.
- 4. HEALTHCARE ANALYTICS MARKET • Real time analytics is carried out on the spot and helps in quick decision making, for instance, clinical decision support software with active knowledge systems use two or more items of patient data to generate case- specific advice. • Batch Analytics retrospectively evaluates past data such as patient records and claims data from an insured population, which helps in predictive modeling and cost control measures.
- 5. PROCESS INVOLVED Project Definition / Business Understanding Exploration / Data Understanding Data Preparation Model Building Deployment
- 6. REQUIREMENT : • Predictive analytics service providers generally start by studying the characteristics of people who have already purchased a product from an insurer, • and then develop a profile — or model — of the kind of person who buys that specific insurance product.
- 7. VARIABLES IN PREDICTIVE ANALYTICS ALGORITHM • Predicting which policy holders (or potential policy holders) will make a claim • And how long it will be until they make the claim. • The more data available on the history of claims • And ‘extraneous’ information about the policy holder
- 8. DATA
- 9. CASE STUDY In this case analysis of hospital data was done to optimize and balance human resources, medication and time spent on each patient to improve clinical outcomes. Fig.1 performs spectral partitioning of the graph that was built using the data from the health-care agency. Understanding the structure of the data and capturing hidden interrelationships helped to improve the existing resource allocation schema. As a result created a model of resource harness that stopped overspending and improved the quality of patient's care.
- 10. Contd………. In this case analysis of patient’s symptoms was taken to predict the development of the disease. Fig.2 demonstrates principal component analysis and support vector machine classifier. Healthcare data analytics allows us to find patterns that help to recognize early stages of the disease and predict its development. This predictive model provides the hospital with an opportunity to control the occurrence of epidemics as well as be more accurate in early diagnosis of the disease.
- 11. MODEL EVALUATION • Entries will be judged by comparing • The predicted number of days a member will spend in the hospital with the actual number of days a member spent in the hospital in DaysInHospital_Y4 (not given to competitors) • Prediction accuracy will be evaluated based on the following metric • The objective function for the model to minimize where 1. i is a member; 2. n is the total number of members; 3. pred is the predicted number of days spent in hospital for member i in the test period; 4. act is the actual number of days spent in hospital for member i in the test period.
- 12. IMPROVING HEALTH CARE DELIVERY • To identify when patients are likely to have a hospital stay • And to direct health care providers to take preventative actions to avoid the hospital stay. • Prediction of product demand, • Options prices, • Turnover likelihood of sales leads.
- 13. APPLICATION Model drug development collaborations that maximize IP and drug discovery. Simulate PRO (Patient Reported Outcomes) for care quality improvement and outcomes. Accelerate time to market for new therapies with strategic portfolio modeling Predict market access and optimize resource allocation for new therapies Predict high risk patients for ACO (accountable care organization) and hospitals. Leverage advanced analytics to reduce hospital readmissions Simulate connected health consumer and recommend technology interventions that drive healthy behavior change. Simulate the financial risks and incentives of emerging reimbursement models for ACO.