A novel artificial intelligence score provides a more accurate prediction of the likelihood that patients with suspected or known coronary artery disease will die within 10 years than established scores used by healthcare professionals worldwide. The research results will be presented today at EuroEcho 2021, a scientific congress of the European Society of Cardiology (ESC).
In contrast to conventional methods, which are based on clinical data, the new score also contains imaging information of the heart, measured by cardiovascular stress magnetic resonance (CMR). “Stress” refers to the fact that an MRI scanner gives patients a drug to mimic the effects of exercise on the heart.
This is the first study to show that machine learning with clinical parameters plus stress CMR can very accurately predict the risk of death. The results suggest that patients with chest pain, shortness of breath, or risk factors for cardiovascular disease should undergo a stress CMR exam and have their scores calculated. This would allow us to provide more intensive follow-up care and advice on exercise, diet, etc. to those most in need.
Dr. Theo Pezel, study author, Johns Hopkins Hospital, Baltimore, USA
Risk stratification is often used in patients with cardiovascular disease or at high risk for cardiovascular disease to customize management aimed at preventing heart attack, stroke, and sudden cardiac death. Traditional calculators use a limited amount of clinical information such as age, gender, smoking status, blood pressure and cholesterol. This study examined the accuracy of machine learning using stress CMR and clinical data to predict 10-year all-cause mortality in patients with suspected or known coronary artery disease, and compared its performance to existing scores.
Dr. Pezel stated, “For clinicians, some of the information we collect from patients does not appear to be relevant for risk stratification. But machine learning can analyze a large number of variables at once and potentially find associations that we didn’t know existed, which improves risk prediction. “
The study included 31,752 patients referred to a center in Paris between 2008 and 2018 for stress CMR, chest pain, shortness of breath on exertion, or high risk of cardiovascular disease, but no symptoms. A high risk was defined as having at least two risk factors such as high blood pressure, diabetes, dyslipidemia and current smoking. The mean age was 64 years and 66% were men. Information was collected on 23 clinical and 11 CMR parameters. Patients were followed up for a median of six years for all-cause death, which was extracted from the French national death register. 2,679 (8.4%) patients died during the follow-up period.
Machine learning was done in two steps. First it was used to select which of the clinical and CMR parameters can and cannot predict death. Second, machine learning was used to build an algorithm based on the key parameters identified in step one, each with a different focus to make the best prediction. Patients were then given a score of 0 (low risk) to 10 (high risk) for the likelihood of death within 10 years.
The machine learning score was able to predict which patients would be alive or dead with an accuracy of 76% (statistically speaking, the area under the curve was 0.76). “That means that the score made the correct prediction in about three out of four patients,” says Dr. Pezel.
Using the same data, the researchers calculated the 10-year risk of total death using established scores (Systematic COronary Risk Evaluation [SCORE], QRISK3 and Framingham Risk Score [FRS]) and a previously derived score with clinical and CMR data (clinical-stressCMR [C-CMR-10]) 2 – none of them used machine learning. The machine learning score had a significantly higher area under the curve for predicting 10-year all-cause mortality compared to the other scores: SCORE = 0.66, QRISK3 = 0.64, FRS = 0.63 and C- CMR-10 = 0.68.
Dr. Pezel said, “Stress CMR is a safe technique that does not use radiation. Our results suggest that combining this imaging information with clinical data in an algorithm created by artificial intelligence could be a useful tool in preventing cardiovascular disease and sudden cardiac death, “in patients with cardiovascular symptoms or risk factors.”
European Society of Cardiology