Previsão de solvência financeira das operadoras de plano de saúde: uma abordagem com machine learning

Abstract

This study proposes the application of predictive models, emphasizing the Random Forest algorithm, to forecast insolvencies in the supplementary health sector. The insolvency of healthcare plan operators can significantly impact millions of beneficiaries. The increasing complexity of regulatory frameworks and the need for predictability in risk management require advanced financial analysis methodologies, making machine learning an efficient solution. The study utilizes data from the National Supplementary Health Agency (ANS) and relevant financial indicators to validate the Random Forest model. Metrics such as accuracy, precision, and recall were analyzed to assess model performance. Results show that the Random Forest algorithm has high predictive capability, reducing false positives. Additionally, the modeling allows continuous adjustments, ensuring more effective forecasts aligned with sector needs. This research highlights the importance of machine learning in financial risk management and suggests, as future work, the implementation of hybrid models and the incorporation of new variables to further enhance prediction accuracy. he applied methodology involved data collection, preprocessing, and the implementation of the Random Forest algorithm for insolvency prediction. The model’s performance was evaluated using metrics such as accuracy, precision, recall, and AUC. The results demonstrate that the developed model effectively differentiates solvent from insolvent operators, offering a robust tool for risk management in the supplementary health sector. This study contributes to the discussion on predictive methodologies for financial insolvency and highlights the potential of Machine Learning in enhancing decision-making processes in regulated financial environments.