Insurance Fraud Detection

Insurance fraud affects not only the financial health of the insurers, but also of innocent people seeking effective insurance coverage. Fraudulent claims are a serious financial burden on insurers and result in higher overall insurance costs. Here are a few examples of the way data analysis can be applied to fight fraud in the insurance industry:

Medical Billing Fraud
- Identify excessive billing — same diagnosis, same procedure
- Identify excessive number of procedures, per day or place of service/day
- Identify mutiple billing of same procedure, same date of service
- Highlight ?upcoding? of procedures. Statistically outlying numbers
- Locate age inappropriate treatments — too young/old for treatment
- Identify duplicate charges on patient bills
- Find doctor and patient with same address
Claims Fraud
- Identify duplicate cliams
- Review submission of multiple/inflated claims
- Find fraudulent family members: i.e., five dependent children born within a two year period.
- Highlight incorrect gender specific treatments
- Flag mutually exclusive procedures: e.g. if appendix removed on 01/10/14, then it would be impossible to have appendicitis on 01/02/15.
- Highlight failure to disclose pre-existing condition (where applicable)

Life Insurance Fraud
- Determine patterns of overpayment of premiums
- Review transaction payments comprising more than one type of payment instrument
- Report multiple accounts to collect funds or payment to beneficiaries
- Report purchase of multiple products in a short period of time
- Review beneficiaries with multiple policies
- Isolate transactions for follow-up where employees are beneficiaries
- Determine agents/brokers with statistically high numbers of claim payouts
- Calculate benefit payments paid for lapsed policies
- Find policy loans that are greater than face value
- Report unauthorized policy changes
- Identify missing, duplicate, void or out-of-sequence check numbers