Large volume of simultaneous transactions from thousands of hospitals all over the country
A leading health insurer uses big data analytics to detect fraudulent claims. Implementing both rule-based and probabilistic algorithms revealed that 12% of claim amounts seemed very suspicious. This translates to potential savings when the same method is incorporated in the decision support system.
Challenges : Flag suspicious transactions among the volume of a transactions that occur every second.
Large volume of simultaneous transactions from thousands of hospitals all over the country
Shortage of qualified medical allied professionals who could confirm the validity of claims
Inevitable blind spots when issuing LOAs (Letter of Approval)
Solution : Embed rule-based and probabilistic algorithms in the authorization decision-support system.
MediLink enhanced business rules with machine learning algorithms to auto-adjudicate claims based on consistency of patterns recognized from past experience.
Collect absolute rules from experts to detect combinations that are impossible
Determine thresholds using probability distribution on different segments of data
Warn users if a transaction parameter violates any of the rules or thresholds.
Impact : Intelligent verification and exception-based human intervention prevent fraud while minimizing processing cost
Faster and broader authentication of transaction data
12% of claims have been flagged as suspicious or potentially fraudulent
Millions in monthly savings
Case Study