Comparison Between Artificial Intelligence-Based Models and Traditional Risk Scores for Predicting Risks in Adult Cardiothoracic Surgery: A Systematic Review.
Tobias K Fuchs, Cameron Jones, Michael Breiner
✦ AI-curated · Sources linked
In 30 seconds
This systematic review evaluated the performance of machine learning (ML) models compared to traditional risk scores in predicting perioperative mortality and adverse events in adult cardiothoracic surgery. Thirteen studies were analyzed, revealing that ML models demonstrated improved or comparable performance, with area under the curve (AUC) differences ranging from 0.006 to 0.42.
Key findings
- ML models showed improved or comparable performance with AUC differences of 0.006-0.42.
- Extreme gradient boosting AUC was 0.96 for postoperative infection.
- Random forest AUC was 0.975 for major adverse events in type A dissection.
- Net reclassification improvement was 0.550.
Why it matters
Understanding the potential of machine learning in risk prediction can enhance decision-making in cardiothoracic surgery, potentially leading to better patient outcomes and more tailored surgical approaches.
What to keep in mind
External validation was limited, and some analyses raised concerns regarding overfitting.