Key result
An XGBoost machine learning model using 12 to 18 preoperative clinical variables accurately predicted 30-day postoperative mortality, achieving an AUROC of 0.941 in external validation.
Why the study?
Accurate prediction of postoperative mortality is important for care, shared decision-making, and resource allocation, motivating a model using manageable clinical inputs and multicenter validation.
Does a machine-learning prediction model using minimal preoperative clinical variables accurately predict 30-day mortality after non-cardiac surgery in adult patients?
Population
454,404 patients over 18 years of age undergoing non-cardiac surgeries across four institutions
Comparison
Logistic regression vs random forest vs XGBoost vs deep neural network methods
Design
Multi-center retrospective cohort study
Follow-up
30-day
Authors
Loading...
May support automated preoperative risk stratification; leaves open whether integration improves outcomes in prospective trials.
Observational (n=454,404)
Yes
Does a machine-learning prediction model using minimal preoperative clinical variables accurately predict 30-day mortality after non-cardiac surgery in adult patients?
Effect estimate: AUROC 0.941 (95% CI 0.938-0.943)
A robust machine-learning model using only 12-18 automatically extractable preoperative clinical variables can accurately predict 30-day mortality after non-cardiac surgery across multiple institutions.
Lee et al. (2022) conducted an observational in Non-cardiac surgery (n=454,404). XGBoost machine learning prediction model vs. ASA-PS classification was evaluated on Prediction of 30-day postoperative mortality (AUROC) (AUROC 0.941, 95% CI 0.938-0.943). An XGBoost machine learning model using 12 to 18 preoperative clinical variables accurately predicted 30-day postoperative mortality, achieving an AUROC of 0.941 in external validation.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: