Tehran University of Medical Sciences Archives of Anesthesiology and Critical Care 2423-5849 0 0 2026 06 22 1575 1575 Atiyeh sadat Sajadi Department of Anesthesia, School of Allied Medical Sciences, Iran University of Medical Sciences, Tehran, Iran. Mehrdad Mesbah Kiaei Department of Anesthesiology and Pain Medicine, School of Medicine, Hasheminejad Kidney Center, Iran University of Medical Science, Tehran, Iran. Raheleh Charmchi Department of Anesthesia, School of Allied Medical Sciences, Iran University of Medical Sciences, Tehran, Iran. Zeinab Barzegar Department of Artificial Intelligence in Medicine, Faculty of Advanced Technologies in Medicine, Iran University of Medical Science, Tehran, Iran. Kimia Khonakdar Department of Anesthesiology, School of Allied Medical Sciences, Mazandaran University of Medical Sciences, Sari, Iran Shahnam Sedigh Maroufi Department of Anesthesia, School of Allied Medical Sciences, Iran University of Medical Sciences, Tehran, Iran. Mahsa Atri Roozbahani Department of Anesthesiology, Mohebe Kosar Hospital, Faculty of Medicine, Iran University of Medical Sciences, Tehran, Iran. Parisa Moradimajd Department of Anesthesia, School of Allied Medical Sciences, Iran University of Medical Sciences, Tehran, Iran. 2026 02 16 2026 02 25 Background: Hypotension, heart rate changes, and postoperative nausea and vomiting (PONV) are common complications of spinal anesthesia in urologic surgeries, which may lead to hemodynamic instability, increased therapeutic interventions, and reduced patient satisfaction. Early identification of high-risk patients can play a crucial role in the prevention and management of these complications. Methods: In this study, data from patients undergoing urologic surgeries under spinal anesthesia were collected, cleaned, and preprocessed. Various machine learning models, including logistic regression, support vector machine, decision tree, random forest, and multilayer perceptron (MLP) neural networks, were trained to predict hypotension, heart rate changes, and PONV. Feature selection was performed using the Boruta algorithm and correlation analysis. Results: The MLP model achieved the strongest predictive performance for hypotension (AUC ≈ 0.86), heart rate changes (AUC ≈ 0.91), and PONV (AUC ≈ 0.87–0.90). The most important predictors were baseline blood pressure and heart rate, ASA status, age, surgical type and duration, and intraoperative medication use. Conclusion: Machine learning models may be useful for identifying patients at high risk before complications develop after spinal anesthesia, offering a basis for building clinical decision support systems. https://aacc.tums.ac.ir/index.php/aacc/article/view/1575