A team of Johns Hopkins researchers has developed an AI-powered stress cardiovascular magnetic resonance (CMR) that predicts the probability of dying in patients with suspected or known coronary artery disease within 10 years better than existing ones used by health professionals globally.
A new artificial intelligence score predicts the chance of individuals with suspected or known coronary artery disease dying within ten years more accurately than existing scores used by health professionals around the world.
The findings were presented today at EuroEcho 2021, the European Society of Cardiology’s scientific convention (ESC).
Unlike old methods based on clinical data, the new score incorporates cardiac imaging data obtained through stress cardiovascular magnetic resonance (CMR). The term “stress” relates to the fact that while in the magnetic resonance imaging scanner, patients are given a medication that mimics the effect 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,” says study author Theo Pezel.
“The findings indicate that patients with chest pain, dyspnoea, or risk factors for cardiovascular disease should undergo a stress CMR exam and have their score calculated. This would enable us to provide more intense follow-up and advice on exercise, diet, and so on to those in greatest need.”
In patients with or at high risk of cardiovascular disease, risk stratification is routinely used to customize care aimed at preventing heart attacks, strokes, and sudden cardiac death. Traditional calculators rely on a limited set of clinical data, such as age, gender, smoking status, blood pressure, and cholesterol.
This study evaluated the performance of machine learning with existing scores in predicting 10-year all-cause death in individuals with suspected or known coronary artery disease using stress CMR and clinical data.
Dr. Pezel further explains: “For clinicians, some information we collect from patients may not seem relevant for risk stratification. But machine learning can analyse a large number of variables simultaneously and may find associations we did not know existed, thereby improving risk prediction.”
Between 2008 and 2018, 31,752 patients were referred to a center in Paris for stress CMR because of chest pain, shortness of breath on exertion, or a high risk of cardiovascular disease but no symptoms.
At least two risk factors, such as hypertension, diabetes, dyslipidemia, and current smoking, were considered high risk. The average age was 64, and 66 percent of the participants were men.
A total of 23 clinical and 11 CMR parameters were gathered. For all-cause death, patients were tracked for a median of six years using data from France’s national death registry. During the study’s follow-up period, 2,679 patients (8.4%) died.
There were two stages to the machine learning process. It was first used to determine which clinical and CMR markers could accurately predict death and which could not. Second, machine learning was used to design an algorithm based on the essential parameters found in step one, giving each a varied level of importance in order to make the best prediction. The likelihood of dying within 10 years was then assigned a score ranging from 0 (low risk) to 10 (high risk).
With 76 percent accuracy, the machine learning score was able to predict whether patients would live or die (in statistical terms, the area under the curve was 0.76).
“This means that in approximately three out of four patients, the score made the correct prediction,” adds Dr. Pezel.
The researchers used the same data to calculate the 10-year risk of all-cause death using established scores (Systematic COronary Risk Evaluation [SCORE], QRISK3, and Framingham Risk Score [FRS]), as well as a previously derived score incorporating clinical and CMR data (clinical-stressCMR [C-CMR-10])2 – none of which used machine learning. When compared to the other scores, the machine learning score exhibited a considerably greater area under the curve for predicting 10-year all-cause mortality: SCORE = 0.66, QRISK3 = 0.64, FRS = 0.63, and C-CMR-10 = 0.68.
Dr. Pezel concludes: “Stress CMR is a safe technique that does not use radiation. 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.”
Image Credit: Getty