Key result
Logistic regression outperforms XGBoost by ~0.04 AUC for predicting in-hospital mortality after cardiac surgery.
Why the study?
The comparative performance and temporal generalizability of machine learning models for surgical risk prediction remain inadequately evaluated.
Does logistic regression perform comparably to complex machine learning models for predicting in-hospital mortality in adult patients undergoing cardiac surgery?
Population
9,956 adult patients undergoing coronary artery bypass grafting and/or valve surgery
Comparison
Logistic regression vs random forest vs XGBoost models
Design
Retrospective cohort study
Authors
Loading...
May support logistic regression for post-cardiac surgery mortality stratification; leaves open ML advantages in prospective cohorts.
Cohort (n=9,956)
No
Does logistic regression perform comparably to complex machine learning models for predicting in-hospital mortality in adult patients undergoing cardiac surgery?
Mean Difference: 0.044 (95% CI 0.002–0.084)
Absolute Event Rate: 0.876% vs 0.831%
Logistic regression provides an interpretable and well-calibrated model for early-admission mortality risk stratification after cardiac surgery, performing comparably to or slightly better than more complex machine learning models.
Zhou et al. (2026) conducted a cohort in Cardiac surgery (CABG and/or valve surgery) (n=9,956). Logistic regression model vs. XGBoost model was evaluated on Temporal AUC for predicting in-hospital mortality (Difference in AUC +0.044, 95% CI 0.002, 0.084). Logistic regression achieved the highest temporal discrimination for predicting in-hospital mortality after cardiac surgery (AUC 0.876), outperforming XGBoost by a small margin (difference +0.044).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: