Machine-learning prediction of impaired outcome in diabetic patients undergoing non-cardiac surgery

To develop interpretable machine-learning models for predicting impaired outcomes in diabetic patients undergoing non-cardiac surgery.

Patients with impaired outcomes were older, had lower body weight, higher ASA class, and greater cardiovascular comorbidity.
AdaBoost model showed the best discrimination with an AUC of 0.82.
Key predictors of risk included prior ischemic stroke, prior myocardial infarction, preoperative creatinine, albumin, inflammatory markers, age, ASA class, heart-rate and systolic blood-pressure summaries, and BUN.

Remove or rephrase to avoid unsupported conclusions.

The follow-up duration for the composite endpoint could not be recovered.
The dataset is institutionally governed and not publicly available.
Further calibration testing and external validation are needed before clinical deployment.

Revise to reflect only findings without editorial interpretation.