Machine Learning Predicts Postoperative Delirium After Cardiac Surgery

A systematic review and meta-analysis published in Journal of Medical Internet Research (2026) analyzed 28 studies through August 30, 2024, covering 80,143 cardiac surgery patients to evaluate machine learning models predicting postoperative delirium. In validation datasets the pooled c-index was 0.805, sensitivity 0.72, and specificity 0.78, with logistic regression commonly used. Authors call for multicenter validation to strengthen risk stratification and targeted prevention.