Machine Learning-Based Insurance Claim Prediction Using Demographic and Health Factors

Abstract

In today’s healthcare environment, the increasing cost of medical treatment, medicines, and hospitalization has created significant financial challenges for individuals and families. Health insurance offers financial security by covering qualified medical expenditures and facilitating access to critical healthcare services. However, the growing volume of insurance claims and variations in demographic and health conditions require efficient and accurate approaches for Claim Prediction and decision-making. This study uses the Insurance Claim Analysis: Demographic and Health Dataset, which has 1,340 records and 11 characteristics, to offer a machine learning-based method for forecasting health insurance claims. The data were preprocessed through missing-value handling, duplicate removal, noise reduction, label encoding, SMOTE-based balancing, Min–Max normalization, and stratified data splitting. The proposed Light Gradient Boosting Machine (Light GBM) model was trained using demographic and health-related features and evaluated using accuracy, precision, recall, and F1-score. The proposed model achieved 98.8% accuracy, 98.9% precision, 98.6% recall, and 98.7% F1-score, demonstrating its effectiveness for insurance claim prediction. The outcomes indicate that the proposed approach can support accurate and efficient analysis of health insurance claims and facilitate data-driven decision-making in insurance applications.