Inteligência artificial, predição informacional e o risco de solvência na saúde suplementar brasileira

Abstract

This thesis presents the results of a research in the context of the Brazilian supplementary health, aiming to investigate the impact of Artificial Intelligence (AI) on the informational prediction of healthcare expenses and the solvency risk of Brazilian Health Insurance Companies (HICs). Through cluster analysis using Fuzzy C-Means and K-means models, real-world databases integrating accounting, healthcare, and sociodemographic data were segmented into risk groups based on the proximity of information within each group. For the informational prediction of healthcare expenses (claims), machine learning algorithms such as K-nearest neighbors (KNN), Random Forest and Extreme Gradient Boosting (XGBoost) were employed. The results reveal that AI models provide an accuracy level between 99,06% and 99,26% in the informational prediction of healthcare expenses for Alpha, Beta, and Gamma HICs, suggesting that AI demonstrates the capacity to positively impact the solvency risk of Brazilian HICs and represents a real possibility for HICs to optimize the inverted financial cycle and resource allocation capability. Based on these findings, guidelines for solvency risk management of HICs were proposed, representing a significant step towards integrating AI technologies into the operational practices of HICs, contributing to a more resilient and sustainable healthcare system. The results have produced contributions to multiple approaches involving economic, social, and corporate governance aspects, demonstrating both theoretical advancement and practical applications with the potential to positively influence the economy, society, environment, and corporate governance in the context of Brazilian supplementary health.