The Predictive Value of Machine Learning for Postoperative Delirium in Cardiac Surgery: Systematic Review and Meta-Analysis - PubMed
4 hours ago
- #cardiac surgery
- #postoperative delirium
- #machine learning
- Postoperative delirium (POD) following cardiac surgery is a severe complication with challenges in early identification.
- Machine learning (ML) is gaining attention for predicting POD risk, but more evidence is needed.
- A systematic review and meta-analysis evaluated ML's performance in predicting POD after cardiac surgery.
- The study analyzed 28 original studies involving 80,143 patients, with 6,326 developing POD.
- ML models showed a c-index of 0.805, sensitivity of 0.72, and specificity of 0.78 in validation datasets.
- Logistic regression was the primary modeling method, with a c-index of 0.773 in validation datasets.
- ML-based tools demonstrate promising performance but require further multicenter studies for robust validation.
- Future research should focus on precise risk stratification and targeted preventive interventions for POD.