Machine Learning Models in the Prediction of Adverse Outcomes in Peripheral Arterial Disease: Meta-Analysis.

BACKGROUND

This systematic review and meta-analysis aimed to evaluate the performance of machine learning (ML) models compared to traditional statistical approaches in predicting adverse outcomes for patients with peripheral arterial disease (PAD), while also assessing the current limitations and challenges in ML model development and validation.

METHODS

A comprehensive search of major databases was conducted for studies published between 2000 and 2025 that applied ML techniques to predict outcomes in PAD patients. Major medical databases (OVID Medline, OVID EMBASE, Cumulated Index in Nursing and Allied Health Literature, and Cochrane Central Register of Controlled Trials) and clinical trial registers were searched. Two reviewers independently screened studies, extracted data, and assessed quality and risk of bias using the Prediction model Risk Of Bias ASsessment Tool. The methodological quality of the studies included in this review was also assessed using an artificial intelligence/ML-specific quality assessment tool. A modified hierarchical summary receiver operating characteristic analysis was performed and diagnostic odds ratios (DORs) were calculated to compare the predictive performance of various ML models and traditional regression methods in the prediction of major adverse cardiovascular events and major adverse limb events (MALEs).

RESULTS

Thirteen studies met the inclusion criteria. Gradient-boosted models demonstrated the highest predictive performance with a DOR of 36.593 (95% confidence interval [CI]: 24.44-54.79), sensitivity of 0.853, and specificity of 0.863 in the prediction of MALE. Even the least performing ML models in each study demonstrate moderate predictive ability with a DOR of 8.616 (95% CI: 5.473-13.564). Traditional regression models consistently underperformed compared to ML approaches, with the lowest DOR of 3.326 (95% CI: 1.814-6.097). The quality assessment revealed a mix of methodological rigor, with 54% of studies rated as low risk of bias and 46% as unclear.

CONCLUSION

ML techniques demonstrate superior predictive power for adverse outcomes in PAD patients compared to traditional regression methods.