Healthcare Provider Fraud Detection Analysis by applying supervised machine learning models

Abstract

This study investigates the useful data mining methods to detect healthcare provider fraud. This research tries to apply multiple supervised learning classification models like Naive Bayes, Random Forest, and Logistic Regression to recognize and predict fraudulent claims by healthcare providers. The evaluation matrices like accuracy, precision, recall, and F1 score are used to evaluate the classifier’s performance. The goal is to predict the potentially fraudulent providers based on their claims and study the patterns to understand the future behavior of the healthcare providers. This study also analyzes and explores different patterns in the data like states with the most reported frauds or the most targeted age group of patients. All the classification models are developed utilizing a synthetic dataset to understand and discover the behavior of potentially fraud providers.