Evaluation of the Financial Distress of Hospitals Through Machine Learning: An Application of AI in Healthcare Industry

Abstract

Due to the intricate nature of hospital structures, the examination of factors contributing to financial distress necessitates more advanced methodologies than conventional approaches. Recent advancements in artificial intelligence, specifically machine learning algorithms, offer alternative means of analyzing patterns in these factors to assess hospital financial distress. This study employs various machine learning algorithms to forecast financial distress, as measured by the Altman Z score, for hospitals in Turkey. Prediction models were constructed using decision trees, random forests, K-nearest neighbors, artificial neural networks, support vector machines, and lasso regression algorithms. The findings indicate that the most effective classifiers for predicting hospital financial distress were lasso regression and random forest. Additionally, financial factors, competition, and socioeconomic development level emerged as significant determinants in forecasting hospital financial distress.