Key result
Machine learning outperforms traditional models in predicting non-cardiac surgery MACE with an AUC of 0.85.
Why the study?
Many risk prediction models have been developed for MACEs in non-cardiac surgery, but their quality and clinical applicability remain unclear.
Do machine learning and traditional prediction models accurately predict major adverse cardiovascular events in patients undergoing non-cardiac surgery?
Meta-Analysis (n=93)
Do machine learning and traditional prediction models accurately predict major adverse cardiovascular events in patients undergoing non-cardiac surgery?
Effect estimate: AUC 0.80 (95% CI 0.76-0.83)
Machine learning-based prediction models demonstrate superior discrimination for perioperative MACEs in non-cardiac surgery compared to traditional methods, though clinical implementation is currently limited by a lack of external validation and high heterogeneity.
Supports ML models for perioperative MACE prediction; leaves open routine use pending external validation and bias mitigation.
Many risk prediction models have been developed for major adverse cardiovascular events (MACEs) in non-cardiac surgery, but their quality and applicability in clinical practice are still unclear. PubMed, Cochrane Library, Embase, and Web of Science were thoroughly searched for studies on perioperative MACEs, encompassing cardiac death, cardiac arrest, acute coronary syndrome, myocardial infarction, heart failure, cardiac arrhythmias, myocardial injury, and stroke in non-cardiac surgery. The Prediction Model Risk of Bias Assessment Tool (PROBAST) was utilized to appraise the risk of bias. The area under the curve (AUC) value was pooled based on the true/false positive and negative numbers, and the pooled effect size was computed. Heterogeneity across studies was judged via the I 2 statistic. After the screening, 93 eligible papers were included. 50 studies performed internal validation, but only 21 studies performed external validation. The overall pooled AUC was 0.80 (0.76–0.83), and the pooled AUC in different subgroups ranged from 0.70 (0.66–0.74) to 0.85 (0.81–0.87). The AUC for new machine learning (ML) was 0.85 (0.81–0.87) and for traditional logistic regression or COX regression was 0.79 (0.75–0.82). Heterogeneity was high overall. 79 studies were graded at high risk of bias. Deeks' funnel plots suggested no publication bias. 19 studies (20.4%) reported calibration values, 15 (16.1%) provided net reclassification improvement (NRI) / integrated discrimination improvement (IDI), and 17 studies (18.3%) performed decision-curve analysis (DCA). 11 studies (11.8%) released user-friendly calculators or apps, and merely one study offered electronic medical record (EMR)-integrated deployment. This study highlights the superior discrimination of published prediction models for MACEs, with ML outperforming traditional methods and SHapley Additive exPlanations (SHAP). Yet, high heterogeneity, scarce external validation, limited calibration reporting, and few user-friendly tools currently restrict their clinical applicability. Not applicable.
No takes yet. Share an insight, caveat, or question.
Liu et al. (2026) conducted a meta-analysis in Major adverse cardiovascular events in non-cardiac surgery (n=93). Risk prediction models for MACEs was evaluated on Overall pooled Area Under the Curve (AUC) for predicting MACEs (AUC 0.80, 95% CI 0.76-0.83). Risk prediction models for major adverse cardiovascular events in non-cardiac surgery demonstrated good discrimination with an overall pooled AUC of 0.80, with machine learning models outperforming traditional methods (AUC 0.85 vs 0.79).
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: