Can An Algorithm Help Diagnose Coronary Artery Disease?

Cardiac Phase Space Tomography Analysis (cPSTA) may provide comparable diagnostic utility to functional tests in assessing coronary artery disease (CAD) without cardiac stress tests, according to a study published in PLOS One.

Researchers from several states performed a prospective multicenter study to develop machine-learned algorithms to assess for CAD and to test the accuracy of the algorithms. Coronary artery disease was defined as one or more 70 percent or higher stenosis, or fractional flow reserve 0.80 or less. A total of 606 subjects participated in the study.

From the collective phase signal data, features were extracted and paired with the known angiographic results. A development set, consisting of signals from 512 subjects, was used for machine learning to determine an algorithm that correlated with significant CAD, the authors wrote. Verification testing of the algorithm was performed utilizing previously untested phase signals from 94 subjects.

The results showed the machine-learned algorithm on blind testing in the verification cohort:

• Sensitivity of 92 percent
• Specificity of 62 percent
• Negative predictive value of 96 percent
• Positive predictive value of 46 percent

The researchers concluded that the initial multicenter results suggested resting cPSTA may have comparable diagnostic utility to functional tests currently used to assess CAD without requiring cardiac stress (exercise or pharmacological) or exposure of the patient to radioactivity.