Enhancing Medical Insurance Pricing Prediction with SHAP-XGBoost for Informed Decision-Making

This research explores the evolving field of predictive modeling in healthcare highlighting the continued interest of insurance companies in using Machine Learning (ML) techniques to improve operational efficiency. The author uses a set of regression based ML models incorporating different versions of Extreme Gradient Boosting (XGBoost) methods to predict medical insurance costs. Furthermore the study utilizes Explainable Artificial Intelligence (XAI) methods, Shapley Additive Explanations (SHAP) to identify and explain the key factors influencing medical insurance premium prices within the dataset. The dataset consists of 986 records from the KAGGLE repository and the models effectiveness is thoroughly assessed using various performance evaluation metrics such as R squared ( \(R^2\) ), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). Additionally a comparison is made between the results generated by XGBoost and Random Forest (RF) models, in determining the features affecting Premium Prices. Despite requiring computational resources the XGBoost model stands out as the top performer overall. The authors aim to offer insights to help policymakers, insurance providers and individuals looking for medical coverage make informed decisions. This will assist them in choosing policies that best suit their needs and preferences.