Q-Wave Myocardial Infarction & Stroke in Patients Managed Interventionally for Left Main Coronary Artery Disease: Prediction by Incorporation of Ensemble Approach into Automated Machine Learning

Abstract

BackgroundPercutaneous coronary intervention (PCI) and coronary artery bypass graft (CABG) are recognised as a couple of time-tested interventions to manage left main coronary artery (LMCA) disease. With Artificial intelligence in general and machine learning in particular exhibiting superior predictive and prognosticative capabilities, when compared with conventional statistical techniques, we explored the current state-of-the-art for automated Machine Learning to develop prediction algorithmic models among LMCA patients managed for post-interventional ‘all-cause mortality, Q-wave myocardial infarction, or stroke’ (MQMIStroke) composite risk and target-vessel revascularization (TVR) with the aim to optimise risk stratification and complication triaging. MethodologyThe study population comprised 2,240 patients suffering from LMCA disease managed via either PCI or CABG as collated in the MAIN-COMPARE registry. The current state of the art (SOTA) for automated Machine Learning (aML) was adopted to develop predictive models using algorithms including Neural Network, eXtreme Gradient Boosting and CatBoost with the employment of hyperparameter tuning. Ensemble approach, which is the amalgamation of two or more algorithmic models to develop such a model which is better than either of its computive components, was superimposed. Macro weighted average Area Under the Receiver Operating Curve (mWA-AUROC) and log loss, along with other parameters, were adopted to assess the predictive discriminative ability of the developed models. ResultsAn ensemble of Decision Tree and CatBoost algorithmic models predicted the post-interventional MQMIStroke composite risk in LMCA disease patients with an mWA-AUROC of 0.85 and a log loss of 0.53 recognizing age as the most influential predictor. (Figure 1a) An ensemble of Light Gradient Boosting Machine, Random Forest and CatBoost algorithmic models predicted post-interventional TVR in such patients with an mWA-AUROC of 0.82 and a log loss of 0.51 recognising the respective intervention as the most influential predictor. (Figure 1b) ConclusionsOur novel approach to developing predictive risk models for post-interventional MQMIStroke composite risk and TVR among LMCA disease patients by exploring automated machine learning provides predictive capabilities which, when incorporated into the respective prognosticative protocols, shall translate into a decrease in the morbidity and mortality associated with this ailment by assisting in risk stratification and complication triaging.These authors contributed equally to data curation, development of methodology, statistical analyses and writing with subsequent reviewing and editing of the abstract draft.These authors, indicated alphabetically, contributed equally to statistical analyses and validation.