Novel Artificial Intelligence (AI)-driven Cardiac Blood Tests

ACC 2019 External Validation Presentation for HART CAD
“Validation of Novel Biomarker-Clinical Score to Predict Presence of Obstructive Coronary Disease”
American College of Cardiology (ACC) Scientific Sessions, New Orleans, Louisiana
March 17, 2019
Highlights
-
Undetected coronary artery disease (CAD) carries significant morbidity and mortality while remaining the leading cause of death.
-
A novel biomarker-based model to predict presence of CAD has potential to non-invasively diagnose CAD and has been shown to increase the accuracy of predicting clinically significant CAD in a prior exploratory study.
-
Study objective was to externally validate a recently developed proteomic assay’s ability to diagnose obstructive CAD (Prevencio HART CAD) in a prospective cohort of patients undergoing diagnostic catheterization.
-
Prospective patients referred to cardiac catheterization at single-center were enrolled and blood specimens were obtained pre-angiography. A total of 500 patients were enrolled over a 6-month period.
-
The score had a diagnostic odds ratio of 4.69 (p<0.001; 3.22, 6.84) and AUC 0.74 compared to original validation cohort (n=278) which had diagnostic odds ratio of 9.74 (p<0.001; 6.05, 16.1) and AUC 0.87.
-
Younger cohorts had AUC 0.81 and may demonstrate higher performance. Patients were referred for angiography for various acute and non-acute indications.
Scroll Down for ACC Presentation