@article{article_1737521, title={Predictive Modeling of 30-Day Readmission Risk: A Machine Learning Approach for Health Management}, journal={Bandırma Onyedi Eylül Üniversitesi Sağlık Bilimleri ve Araştırmaları Dergisi}, volume={8}, pages={441–455}, year={2026}, DOI={10.46413/boneyusbad.1737521}, url={https://izlik.org/JA39XD32TA}, author={Doğan, Alican}, keywords={Yeniden yatış riski, Öngörüsel modelleme, Risk yönetimi göstergeleri, Klinik karar destek, Sağlık yönetimi, Makine öğrenmesi}, abstract={Aim: This study aims to develop and evaluate predictive models capable of identifying patients at risk of 30-day readmission using structured inpatient data. Material and Method: The analysis was conducted on a fully synthetic dataset designed to reflect the complexity of real-world clinical data while ensuring the protection of personal health information. The dataset includes key features relevant to inpatient care, such as patient demographics, diagnosis codes, laboratory abnormalities, procedural counts, length of stay, and risk scores. A comprehensive preprocessing phase involving label encoding, data balancing, and normalization is completed. In the data preprocessing stage, categorical variables were encoded, class imbalance was corrected using the SMOTE method, and Logistic Regression, Random Forest, and XGBoost models were applied. Results: According to the results, the Random Forest model demonstrated the highest performance, achieving over 92% accuracy and an AUC value of 0.95. Feature significance analysis revealed that risk score, length of stay, and number of procedures were the most effective variables in determining the risk of readmission. The findings indicate that predictive analytics approaches can make significant contributions to early intervention planning and resource optimization in healthcare management. Conclusion: The proposed modeling approach may support early intervention planning and hospital resource optimization while promoting data-driven strategies to reduce preventable readmissions.}, number={2}