CMS Medicare Fraud Detection
-
Jumprice Technologies Inc.
Healthcare is a major industry in the U.S. with both private and government run programs. Healthcare fraud is a main problem that causes substantial monetary loss in Medicare/Medicaid and insurance industry. The impact of healthcare fraud is estimated to be between 3% to 10% of the nation’s total healthcare spending and continuing to adversely impact the Medicare program and its beneficiaries.This model is an innovative data science tool that predicts fraud in the medical insurance industry using anomaly analysis and geo-demographic metrics. Comprehensive machine learning model is detecting fraud pattern based on the different features: Service Providers (Doctors, Pharmacies), Insurance subscribers (patients), Geo-demographic and commonly abuse drugs prescriptions.
Following open datasets are used to build the model: • Part D Prescriber dataset • Excluded (LEIE) dataset • Payment Received dataset • FDA Drug ingredients dataset • Others: Part D Opioid Prescriber Summary. This tool can be used by insurance providers, health care providers, patients, pharmacy, and doctors.
Input variables : The number of different drugs prescribed, Medical provider’s specialty, Number of procedures/services the provider performed
Output Variables : Medicare claims are fraud or not
Metrics to Monitor
|
Statistical
|
:
|
Somers D |
Accuracy |
Precision and Recall |
Confusion Matrix |
F1 Score |
Roc and Auc |
Prevalence |
Detection Rate |
Balanced Accuracy |
Cohen's Kappa |
Concordance |
Gini Coefficent |
KS Statistic |
Youden's J Index
|
|
Business
|
:
|
Claims Processed |
$ Saved |
FWA Rate |
FWA by CPT |
FWA by Provider
|
|
Infrastructure
|
:
|
Log Bytes |
Logging/User/IAMPolicy |
Logging/User/VPN |
CPU Utilization |
Memory Usage |
Error Count |
Prediction Count |
Prediction Latencies |
Private Endpoint Prediction Latencies |
Private Endpoint Response Count
|
Visit Model :
github.com