Background: Fertile women represent a socially and medically significant patient group, yet little research has examined their rehospitalization behavior and financial burden in clinical settings. This study develops predictive and explainable artificial intelligence for rehospitalization and medical costs among reproductive-age orthopedic patients. Methods: Electronic health records of 83 women (aged 15–49) at a major university hospital in Korea were analyzed. Six machine learning models were developed, and model performance was assessed using accuracy, the area under the curve, the root mean square error and its scaling invariant divided by the interquartile range (RMSE/IQR). Shapley Additive Explanations were applied to interpret predictors of rehospitalization. Additional analyses explored determinants of patients’ total and uncovered medical costs. Results: The random forest outperformed other models in predicting rehospitalization (area under the curve 0.92 vs. 0.73 for logistic regression). Key predictors included major disease, systolic blood pressure, platelet count, age, and treatment costs. The random forest also yielded lower error rates than linear regression in forecasting patients’ costs (e.g., RMSE/IQR for total cost: 1.05 vs. 1.14). Several factors—such as blood pressure, pulse, and hematocrit—were influential for both rehospitalization and costs. Conclusions: Predictive and explainable artificial intelligence can support medical centers in anticipating the rehospitalization and financial burden of fertile women. By integrating medical and socioeconomic determinants, hospitals may design strategies that enhance patient rehospitalization while addressing broader societal priorities in women’s health.
Kwang-Sig et al. (Sat,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: