Can machine learning unravel unsuspected, clinically important factors predictive of long-term mortality in complex coronary artery disease? A call for ‘big data’

Abstract

Aims Risk stratification and individual risk prediction plays a key role in making treatment decisions in patients with complex coronary artery disease (CAD). The aim of this study was to assess whether machine learning (ML) algorithms can improve discriminative ability and identify unsuspected, but potentially important, factors in the prediction of long-term mortality following percutaneous coronary intervention (PCI) or coronary artery bypass grafting (CABG) in patients with complex CAD. Methods and results To predict long-term mortality, the ML algorisms were applied to the SYNTAXES database with 75 pre-procedural variables including demographic and clinical factors, blood sampling, imaging, and patient-reported outcomes. The discriminative ability and feature importance of the ML model was assessed in the derivation cohort of the SYNTAXES trial using a 10-fold cross validation approach. The ML model showed an acceptable discrimination (AUC = 0.76) in cross-validation. C-reactive protein, patient-reported pre-procedural mental status, Gamma-glutamyl transferase, and HbA1c were identified as important variables predicting 10-year mortality. Conclusions ML algorithms disclosed unsuspected, but potentially important prognostic factors of very long-term mortality among patients with CAD. A “mega-analysis” based on large randomized or non-randomized data, so called “BIG Data”, may be warranted to confirm these findings.

10 Years of SYNTAX

“…Interestingly, ML algorithms identified unsuspected, but potentially important prognostic factors predictive of 10-year mortality among patients with CAD such as CRP, patient-reported preprocedural physical and mental status, Gamma-glutamyl transferase, and hemoglobin. 28 Our ML model, which was built using clinical factors, blood sampling, imaging parameters, and patient-reported outcomes, demonstrated that those unconventional heterogeneous variables may have similar importance to conventional variables in predicting 10-year mortality. These variables are not always collected in clinical trials but may be available in large health care datasets.…”

“…Recently, ML has emerged as a novel approach for developing risk models predictive of clinical outcomes. 28 ML and penalized regression can handle large numbers of variables and diverse parameters. Rousset et al 29 incorporated all 428 variables in the FOURIER (Further Cardiovascular Outcomes Research With PCSK9 Inhibition in Subjects With Elevated Risk) trial to an ML model and showed that this performed better than linear regression.…”

“…The high-risk group classified by the ML model had significantly higher observed 10-year mortality than the intermediate (47.3% vs 18.7%, HR: 2.43; 95% CI: 2.12-2.78; P < 0.001) and low-risk groups (47.3% vs 10.7%, HR: 3.17; 95% CI: 2.55-3.95; P < 0.001) ( Figure 16A ). 28 The 10 most important variables to predict 10-year mortality from the best performing model are shown in Figure 16B . Interestingly, ML algorithms identified unsuspected, but potentially important prognostic factors predictive of 10-year mortality among patients with CAD such as CRP, patient-reported preprocedural physical and mental status, Gamma-glutamyl transferase, and hemoglobin.…”

“…Reproduced with permission from Ninomiya et al. 28 AUC = area under the curve; CRP = C-reactive protein; LVEF = left ventricular ejection fraction; PVD = peripheral vascular disease; SF-36 MCS = 36-Item Short Form Survey Mental Component Summary. …”