An Advanced Machine Learning Models Design for Fraud Identification in Healthcare Insurance
Sriram Pabbineedi, Mitra Penmetsa, Jayakeshav Reddy Bhumireddy, Rajiv Chalasani, Mukund Sai Vikram Tyagadurgam, Venkataswamy Naidu Gangineni · International Journal of Artificial Intelligence Data Science and Machine Learning · 2021
Healthcare fraud threatens the interests of both healthcare facilities and the patients that they serve. In turn, such actions usually cause enormous financial damage and compromise the trustworthiness of healthcare systems. The proposed study intends, by adding a machine learning framework, to counter the issue of healthcare fraud. The systematic analysis of healthcare data shows patterns and anomalies which can be used to identify fraudulent behavior with more precision and speed. Based on the XG boost algorithm, this work describes a machine learning approach to irregularity detection and health insurance premium estimation. XG Boost algorithm was evaluated using the standardized performance measures such as R², MAE, RMSE and MAPE. Model’s 89. 47% R² coupled with MAE of 1. XGBoost showed more predictive performance than Random Forest and Genetic Support Vector Machines (GSVMs) when compared. Support for generalization of the model was also offered when learning curves and prediction error plots were considered. From these results it is evident that XGBoost is a reliable approach for detecting insurance fraud and pricing within structured healthcare settings