Key result
Machine learning shows comparable predictive performance to existing clinical risk scores in cardiac surgery.
Why the study?
Machine learning is developing fast within medicine and perioperative care, prompting a review of the extent and potential limitations of its implementation in perioperative anesthetic care specifically in cardiac surgery patients.
Do machine learning models improve perioperative risk prediction and anesthetic management compared to conventional statistical methods in cardiac surgery patients?
Comparison
ML applications vs conventional statistical methods
Authors
Loading...
Should not yet replace clinical scores for cardiac surgery risk prediction; leaves open whether dynamic machine learning models improve perioperative management in prospective studies.
Do machine learning models improve perioperative risk prediction and anesthetic management compared to conventional statistical methods in cardiac surgery patients?
Machine learning models in cardiac surgery currently perform similarly to conventional risk scores for static prediction, but offer significant potential when applied to real-time dynamic perioperative data.
Rellum et al. (2021) conducted a review in Cardiac surgery (n=46). Machine learning models vs. Conventional statistical methods and clinical scores was evaluated on Model performance (AUC/C-index) for event/risk prediction, hemodynamic monitoring, and echocardiography automation. Machine learning models in cardiac surgery generally yield comparable predictive outcomes to existing clinical scores, though models incorporating real-time dynamic variables show promising improvements.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: