AI Assessment of PET/CT May Hold 'Major Promise' for Predicting Heart Attack Risk

 But WHOM  is going to do the research on predicting stroke risk from this?

AI Assessment of PET/CT May Hold 'Major Promise' for Predicting Heart Attack Risk

New research suggests an emerging machine learning model that combines findings from advanced imaging with clinical data may improve risk stratification in people with coronary artery disease.

In a recently published study in the Journal of Nuclear Medicine, researchers found that artificial intelligence (AI)-enabled assessment of combined findings from positron emission tomography (PET), computed tomography (CT) and clinical data may provide “substantial improvement” in predicting heart attack risk for people with coronary artery disease.

In the study involving 293 people with coronary artery disease, the authors noted that 22 study participants suffered a myocardial infarction during the 53-month follow-up period. Employing a machine-learning model, researchers assessed the combination of ¹⁸F-sodium fluoride (¹⁸F-NaF) PET and quantitative plaque analysis via CT angiography along with clinical findings and compared it to these individual diagnostic measures for identifying myocardial infarction risk.

The study findings were as follows:

• Clinical characteristics: (c-statistic 0.64, 95% CI 0.53-0.76)

• Quantitative plaque analysis: (c-statistic 0.72, 95% CI 0.60-0.84)

• ¹⁸F-NaF coronary uptake: (c-statistic 0.76, 95% CI 0.68-0.83)

• Combination of all measures: (c-statistic 0.85, 95% CI 0.79-0.91)

For patients with advanced coronary atherosclerosis, the study authors maintained that predicting risk for myocardial infarction doesn’t depend on cardiovascular risk scores, stenosis severity or CT calcium scoring. They said the study showed that disease activity assessment via ¹⁸F-NaF PET and analysis of plaque type and burden by coronary CT angiography are primary determinants in risk stratification of this patient population.

“Our machine-learning approach has overcome the challenges posed by collinearity of these variables and, for the first time, has demonstrated that this information is complementary and additive with the combination of both providing the most robust outcome prediction,” wrote Piotr Slomka, PhD, FACC, FASNC, FCCPM, a professor of medicine and cardiology within the Division of Artificial Intelligence in Medicine at Cedars-Sinai Medical Center in Los Angeles, and colleagues. “If confirmed in future studies, this comprehensive approach holds major promise in refining risk stratification of patients with established coronary artery disease, a population for which such prediction is currently challenging.”

Conceding a limited number of patients and myocardial infarction events in this study, the authors noted larger studies are necessary to validate the findings. They said one ongoing prospective study is assessing the aforementioned modalities for the prediction of recurrent events in patients who had a recent heart attack and multivessel disease.