A Predictive Analytics Approach for Health Insurance Cost Modeling in Pharmaceutical Healthcare

Abstract

Health insurance policies provide financial support to individuals by covering medical expenses and reducing the financial burden associated with diseases, injuries, and healthcare treatments. Several demographic and health-related factors influence the cost of health insurance, making accurate cost prediction an important task for insurance providers and policyholders. Early prediction of health insurance expenses can assist in determining suitable coverage plans and improving financial decision-making. In this study, a supervised machine learning-based approach is proposed for Health Insurance Cost Prediction using the Medical Insurance Cost Dataset. Data preprocessing techniques, including missing value handling, duplicate removal, outlier detection, one-hot encoding, and feature standardization, were applied to enhance data quality and model performance. Four machine learning algorithms, namely Random Forest (RF), AdaBoost, LightGBM, and Extreme Gradient Boosting (XGBoost), were implemented and compared. Experimental results reveal that the proposed XGBoost model achieved superior performance with an R-Square (R²) value of 99% and a Mean Absolute Error (MAE) of 721.05, outperforming the other models. The findings demonstrate the effectiveness of XGBoost in capturing complex relationships among insurance-related factors and generating highly accurate cost predictions. Therefore, the proposed model can serve as a reliable decision-support tool for health insurance pricing, risk assessment, and healthcare financial planning.