Comparison of machine learning and traditional methods for prediction of adverse clinical events after percutaneous coronary intervention.

Abstract

Background: Accurate prediction of clinical outcomes following percutaneous coronary intervention (PCI) is essential for mitigating risk and periprocedural planning. Traditional risk models have demonstrated modest predictive value. Machine learning (ML) models offer an alternative risk stratification that may provide improved predictive accuracy. We performed a systematic review and meta-analysis comparing ML models and traditional risk scores for prediction of clinical outcomes after PCI. Methods: This study was reported according to PRISMA guidelines. PubMed, EMBASE, Web of Science and Cochrane databases were searched until 1st November, 2023 for studies comparing ML models with traditional statistical methods for event prediction after PCI. The primary outcome was comparative discrimination measured by C-statistics with 95% confidence intervals between ML models and traditional methods in estimating the risk of all-cause mortality, major bleeding and the composite outcome major adverse cardiovascular events (MACE). Results: Thirty-four models were included across 13 observational studies (4105916 patients). The summary C-statistic all ML models across all clinical endpoints was 0.81 (95% CI, 0.78-0.85), compared to traditional methods 0.79 (95% CI, 0.74-0.85). The difference in C-statistic was 0.02 (95% CI, 0.002-0.04, p=0.56). Of included studies, only 1 model was externally validated and 10 models were calibrated. Conclusion: ML models did not outperform traditional risk scores in the discrimination of all-cause mortality, major bleeding or MACE following PCI. While integration of ML algorithms into electronic healthcare systems may improve periprocedural risk stratification, immediate implementation in the clinical setting remains uncertain. Further research is required to overcome methodological and validation limitations.