Machine learning used to generate risk score based on clinical data and measurements from stress cardiovascular magnetic resonance
THURSDAY, Dec. 16, 2021 (HealthDay News) — A machine learning (ML) score based on clinical data and stress cardiovascular magnetic resonance (CMR) measurements can predict 10-year all-cause mortality in patients with known or suspected coronary artery disease (CAD), according to a study presented at EuroEcho 2021, a scientific congress of the European Society of Cardiology, held from Dec. 9 to 11 in Berlin.
Theo Pezel, M.D., from the Johns Hopkins Hospital in Baltimore, and colleagues examined the feasibility and accuracy of ML using stress CMR and clinical data to predict 10-year all-cause mortality in patients with known or suspected CAD. All consecutive patients referred for stress CMR between 2008 and 2018 were included, with a median follow-up of 6.0 years. Overall, 23 clinical and 11 stress CMR parameters were examined.
The researchers found that 8.4 percent of the 31,752 consecutive patients died with 206,453 patient-years of follow-up. Compared with the clinical-stress CMR-10 (C-CMR-10) score, European Society of Cardiology (ESC)-score, QRISK3-score, Framingham Risk Score (FRS), and stress CMR data alone, the ML score exhibited a higher area under the curve for prediction of 10-year all-cause mortality (0.76 for ML versus 0.68 for C-CMR-10, 0.66 for ESC, 0.64 for QRISK3, 0.63 for FRS, 0.66 for extent of inducible ischemia, and 0.65 for extent of late gadolinium enhancement).
“Our findings suggest that combining this imaging information with clinical data in an algorithm produced by artificial intelligence might be a useful tool to help prevent cardiovascular disease and sudden cardiac death in patients with cardiovascular symptoms or risk factors,” Pezel said in a statement.
Copyright © 2021 HealthDay. All rights reserved.
