Whether you pluck the petals from a daisy or gaze into a crystal ball, the traditional means of predicting and influencing future events has always carried a high degree of risk. Fortunately, when it comes to analyzing claims data, there are proven methodologies —predictive analytics — that states can use to help assess the validity of Medicaid claims.
Predictive analytics encompasses a series of approaches used to compare claims across provider peer groups and validated benchmarks. Claims failing to meet expected patterns in type and frequency of visits, diagnoses, prescriptions and other factors are flagged for investigation.
You probably use predictive analytics to detect fraud, waste and abuse after payment has been made. But recovering misspent funds isn’t easy. It takes time, effort and even more money, and it’s never guaranteed. By applying predictive analytics to the front end of the claims cycle, before payment is made, you can avoid making improper payments to begin with.
Predictive modeling, risk scoring, trends analysis and link analysis are just a few approaches falling under predictive analytics.
- Predictive modeling uses a variety of methods to analyze relevant historical data to create a statistical model of future behavior.
- Risk scoring, like a consumer credit score, assesses providers based on billing, claims and other public and private data.
- Trend analysis examines behavior or activity over time to identify trends and project future direction.
- Link analysis helps identify unusual or hidden relationships to expose fraudulent providers.
For a quick guide to these and other demonstrated approaches to cost avoidance, please read “Predictive analytics to reduce fraud, waste and abuse — best practices.”
— Jeremy Hill
About the author
Jeremy Hill is Optum program integrity director.