– Machine learning tools could predict patients’ long-term risk of heart attacks and cardiac deaths better than standard methods used by cardiologists, according to a study published in Cardiovascular Research.
Coronary artery calcium (CAC) scoring with non-contrast computed tomography (CT) is increasingly used for cardiovascular risk stratification and provides a measure of coronary atherosclerotic burden, or the buildup of cholesterol and other substances on the artery walls.
The researchers noted that several studies have found that the total CAC measured by non-contrast CT predicts cardiovascular events beyond standard cardiovascular risk factors. However, researchers have not yet used machine learning risk scores after CAC scanning for the prediction of future cardiovascular events.
Researchers set out to evaluate a machine learning approach to predicting cardiovascular events. The team studied subjects from the imaging arm of a prospective, randomized clinical trial, who underwent CAC scoring with available cardiac CT scans and long-term follow-up.
Participants were asymptomatic, middle-aged subjects with cardiovascular risk factors, but no known coronary artery disease.
The team used machine learning to evaluate the risk of myocardial infarction and cardiac death in participants, and then compared the predictions with the actual experiences of the subjects over fifteen years. Researchers used a questionnaire to learn about subjects’ cardiovascular risk factors, diets, exercise habits, and marital status.
Over the span of fifteen years, 76 of the 1,912 subjects studied presented an event of myocardial infarction or cardiac death. The results showed that the subjects’ predicted machine learning scores aligned accurately with the actual distribution of observed events.
The atherosclerotic cardiovascular risk disease score, the standard clinical risk assessment used by cardiologists, overestimated the risk of events in the higher risk categories. The machine learning algorithm did not. In unadjusted analysis, high predicted machine learning risk was significantly associated with a higher risk of a cardiac event.
“In this prospective trial, machine learning demonstrated high performance in risk assessment for myocardial infarction and cardiac death in asymptomatic subjects,” the researchers stated.
“By objectively combining clinical data and quantitative CT measures, machine learning provided significantly superior risk prediction compared with the CAC score. These promising results suggest that machine learning has a potential for clinical implementation to improve risk assessment.”
The team noted that an important advantage of the study was the fact that the machine learning score integrated clinical data and quantitative CT assessment after CAC scoring can significantly improve risk assessment.
“In our study, variable importance showed that both clinical and imaging-based parameters were important to provide an optimal risk estimation,” the researchers said.
“As expected, age, atherosclerotic cardiovascular disease (ASCVD) risk score and calcium-related information were the highest contributor to machine learning prediction, but other parameters were also important to provide more accurate estimation. For example, systolic blood pressure, even though it is already included in the ASCVD risk score, still appeared as an independent strong covariate for the risk prediction.”
Going forward, researchers expect that machine learning will help providers assess patient risk in real time, enabling more personalized care and better outcomes.
“With constant growth of artificial intelligence across various disciplines, especially in cardiology, robust cardiac risk assessment will benefit from quantification of automated imaging biomarkers, increasing the relevant information available for clinical decision making,” the team concluded.
Credit: Google News
