Improved ML Model and XAI Methods for Medical Insurance Premium

This paper explores the integration of a machine learning model within a web-based application for the real-time prediction of healthcare insurance premiums based on individual health data. Utilizing a RandomForest regression model initialized with a set of hyperparameters. To optimize the model's performance, a GridSearchCV is employed. The model claims an R2 score of 0.901, the study aims to enhance the accuracy and efficiency of insurance premium estimations, offering a scalable solution that can significantly improve decision-making processes in the insurance sector. The model was trained and evaluated on a dataset comprising health attributes from 986 individuals, achieving robust performance metrics. Subsequently, the model was embedded into a Flask web application, providing a user-friendly interface for instant premium estimations. The deployment on a server facilitated the efficient handling of multiple user requests with minimal latency. This work also implements the Methods in Explainable AI (XAi) designed to clarify how algorithms make decisions, ensuring that patients, healthcare managers, and insurers can all understand and trust these processes. This clarity is vital because it helps reveal and address any biases that may be concealed within the inherently opaque operations of these algorithms.