Main article

Tarasovskaya Nataliya
Margulan University, Pavlodar, Kazakhstan
ziiyuu@gmail.com
Siska Nia Irasanti
Department of Public Health, Faculty of Medicine,Universitas Islam Bandung, Indonesia
ziiyuu@gmail.com
Myat Thida Win
Internal Medicine Department, Faculty of Medicine, University of Cyberjaya, Malaysia
ziiyuu@gmail.com

Abstract

This study explored the application of machine learning models in predicting the readmission rate of postoperative patients. We used random forests and Gradient Boosting Machines (GBM) to evaluate their performance in predicting the risk of readmission. The results showed that although these models showed good recall and were able to effectively identify high-risk patients, there was still a trade-off between recall and precision. Key clinical characteristics such as postoperative complications, age, type of surgery, and length of stay were identified as important predictors of readmission risk. Although these models have achieved encouraging results, they still face challenges in clinical application, especially in terms of interpretability and fairness. Future research should focus on enhancing the interpretability of the models and expanding the dataset and introducing new algorithms that can enhance the prediction performance, ultimately contributing to better management of patients and optimizing the allocation of medical resources. 

Article details

How to Cite

Nataliya, T., Irasanti, S. N., & Win, M. T. (2025). Explainable AI Models for Predicting Postoperative Readmission: Enhancing Surgical Aftercare with Interpretable Intelligence. Journal of AI in Healthcare and Biomedical Engineering, 3(1), 1-16. https://doi.org/10.63646/jaihbe.2025.030101