Impact Solution Challenge 1 US STATE AUDITOR Identify and Prevent Fraud Problem type: Anomaly detection Universal relevance: Fraud detection is usually manual and labor-intensive. At scale it can be a tricky problem to tackle. However, if you apply AI properly you can both save time and be more effective. • Healthcare fraud can be challenging to detect - Even for intelligent medical providers & patients - Fraudsters well versed in remaining undetected • Using a manual process to catch fraud wasn’t working - Random samples selected for inspection - Less than 5% of all transactions were inspected - Time-consuming & limited resources • Focused on specific behavioral patterns • Easily prototyped & tested effectiveness of ML • Supervised learning deployed: - Scan and flag high-risk fraud ca

Join for free to read