AI predicts heart attacks better than existing risk models
Machine learning (ML) can help healthcare providers predict heart disease—including heart attacks—better than other popular risk models, according to new research published in Radiology.
Lead author Kevin M. Johnson, MD, and colleagues worked to craft an ML system that could extract additional details out of coronary computed tomography angiography (CCTA) images “for a more comprehensive prognostic picture.”
“Starting from the ground up, I took imaging features from the coronary CT,” Johnson, an associate professor of radiology and biomedical imaging at the Yale School of Medicine in New Haven, Connecticut, said in a prepared statement. “Each patient had 64 of these features and I fed them into a machine learning algorithm. The algorithm is able to pull out the patterns in the data and predict that patients with certain patterns are more likely to have an adverse event like a heart attack than patients with other patterns.”
Johnson et al. compared this new ML system with the coronary artery disease reporting and data system (CAD-RADS), a decision-making tool currently used to summarize a patient’s CCTA results. Data from more than 6,000 patients was used for the study, following the patients for an average of nine years after they underwent CCTA. A total of 380 patients died during that time, 70 died from coronary artery disease and 43 reported experiencing a heart attack.
Overall, the ML system had an area under the ROC curve (AUC) of 0.77 for predicting all-cause mortality, higher than the CAD-RADS AUC of 0.72. The ML system also had a higher AUC for coronary artery heart disease deaths (0.85 vs 0.79).
“The risk estimate that you get from doing the machine learning version of the model is more accurate than the risk estimate you’re going to get if you rely on CAD-RADS,” Johnson said in the same statement. “Both methods perform better than just using the Framingham risk estimate. This shows the value of looking at the coronary arteries to better estimate people’s risk.”
Johnson is still hard at work researching this topic; he said in the statement that he is already working on taking the ML system to another level by including risk factors not related to imaging results.
“If you add people’s ages and particulars like smoking, diabetes and hypertension, that should increase the overall power of the method and improve the overall results,” he said.
Michael has more than 18 years of experience as a professional writer and editor. He has written at length about cardiology, radiology, artificial intelligence and other key healthcare topics.