Validation of Health Insurance Customers Using XGBoost Modeling

Abstract

Given the inadequate risk assessment of policyholders in the insurance industry, particularly in health insurance, there is a significant emphasis on the validation modeling for customers’ creditworthiness. Therefore, the current study aimed to provide the modeling for health insurance customers validation, with a specific focus on individuals covered by health insurance, particularly employees of the East Iran Oil Company. In this study, method XGBoost using machine learning were employed as the top artificial intelligence methods for customer validation.
Notably, the validation process identified approximately 1.78% of the population as "unhealthy." This seemingly small group accounts for a disproportionately high 17.47% of the company's total health insurance claims. After the training process, the designed model was evaluated using various metrics, including standard metrics, such as Accuracy, Precision, Recall, and F-measure, each examining specific features of the model. The values of these metrics were 0.999, 0.992, 1, and 0.996, respectively. These values were indicative of the very high accuracy, precision, and efficiency of the model. This type of validation model is one of the most practical modeling approaches that insurance companies can use to validate their customers in order to pay an insurance premium in proportion to the level of risk.