C2S Technologies worked with customer to solve client problem
Healthcare – Fraud Detection
Enabled Capabilities
Neural networks, genetic algorithms, nearest neighbor methods were enabled to detect prescription fraud, Upcoding (claims for medical procedure which is more expensive or not performed at all).
Outcome
6% prescriptions claims are the patients who do not exists, which save significantly around $500K.
Saved avg $2M per year in up coding claims which are not mandatory and did not perform.
Customer
Impact Created
Found 6% of prescriptions claims from the patients who do not exists, and 8% upcoding claims which are expensive and 3% of them include did not perform services at all.
Summary
Customer receives medical claims for reimbursement. All claims are coming from insurance companies, clinics, and hospitals & Pharmacies. Claims were increasing day by day and there is was no system to measure a claim is a genuine or fraud. Customer has decided to do data mining to aid the prevention and early detection of medical insurance fraud, fraudulent claims.
Challenge
Customer receives thousands of claims every day for reimbursement from insurance companies, clinics, and hospitals & Pharmacies. Every claim has been approved no prior background whether it is genuine claim or not. Claims cost exceeding given budget. To prevent the leakage, customer has to have a right system in place to aid the prevention and early detection of medical insurance fraud, fraudulent claims, but traditional software solutions cannot solve these kind of problems.
Approach
C2S Technologies worked with customer to solve client problem.
We built machine learning models for following techniques to detect fraud
- Gap Testing
- Classification
- Summing of numeric values
- Validating entry dates
- Calculation of statistical parameters
Getting Started Requires an understanding of
- The areas in which fraud can occur
- What fraudulent activity would look like in the data
- What data sources are required to test for indicators of fraud
