Machine-learning score using stress CMR for death prediction in patients with suspected or known CAD

In patients with suspected or known coronary artery disease (CAD), traditional prognostic risk assessment is based upon a limited selection of clinical and imaging findings. Machine learning (ML) methods can take into account a greater number and complexity of variables.

To investigate the feasibility and accuracy of ML using stress CMR and clinical data to predict 10-year all-cause mortality in patients with suspected or known CAD, and compared its performance to existing clinical or CMR scores.

Between 2008 and 2018, a retrospective cohort study with a median follow-up of 6.0 years (interquartile range: 5.0–8.0) included all consecutive patients referred for stress CMR. Twenty-three clinical and 11 stress CMR parameters were evaluated. Machine learning involved automated feature selection by random survival forest, model building with a multiple fractional polynomial algorithm, and 5 repetitions of 10-fold stratified cross-validation. The primary outcome was all-cause death based on the electronic National Death Registry. The external validation cohort of the ML score was performed in another center.

Of 31,752 consecutive patients (mean age 63.7±12.1 years and 65.7% males), 2,679 (8.4%) died with 206,453 patient-years of follow-up. ML score (ranging 0 to 10 points) exhibited a higher area-under-the-curve compared with C-CMR-10-score, ESC-score, QRISK3-score, FRS and stress CMR data alone for prediction of 10-year all-cause mortality (ML: 0.76 vs. C-CMR-10-score: 0.68, ESC-score: 0.66, QRISK3-score: 0.64, FRS: 0.63, extent of inducible ischemia: 0.66, extent of LGE: 0.65, all p<0.001). The ML score exhibited also a good area-under-the-curve in the external cohort (AUC: 0.75).

The ML score including clinical and stress CMR data exhibited a higher prognostic value to predict 10-year death compared with all traditional clinical or CMR scores.

Type of funding sources: None.