Coronary Artery Disease I: Introduction
Coronary Artery Disease II: Pathophysiology
Imaging Studies for Cardiovascular System VI: Calcium -Scoring CT
Acute Coronary Syndrome III: Diagnostic Studies
You might also read
Articles linked to this work by shared authors, journal, and citation graph.
Updated: Aug 28, 2025

Signal Acquisition, Score Interpretation, and Economics of a Non-Invasive Point-of-Care Test for Coronary Artery Disease
Published on: August 9, 2024
Thomas Stuckey1, Frederick Meine2, Thomas McMinn3
1Cone Health Heart and Vascular Center, Greensboro, NC, United States.
This study introduces a new machine learning algorithm for detecting coronary artery disease. The algorithm uses voltage and photoplethysmographic signals collected in an office setting. It does not require radiation or patient stress. The algorithm was validated against standard diagnostic tests. At certain thresholds, it matches the accuracy of these tests. The system offers advantages in accessibility and patient comfort. It may serve as a front-line diagnostic tool. The algorithm provides a non-invasive alternative to current diagnostic methods.
Area of Science:
Background:
Current diagnostic tests for coronary artery disease vary widely in performance and require specialized equipment, radiation exposure, or patient stress. Standard methods like SPECT imaging involve high costs and logistical challenges. Prior research has shown limitations in accessibility and patient comfort. This gap motivated the development of a non-invasive alternative. No prior work had resolved the need for a radiation-free, stress-free test. The need for a simpler, office-based solution remains unmet. Existing methods lack portability and ease of use. This paper introduces a novel machine learning approach to address these limitations.
Purpose Of The Study:
The aim of the study was to develop and validate a machine learning algorithm for identifying functionally significant coronary artery disease. The algorithm is designed to operate in an office setting without requiring radiation or patient stress. The study focused on improving accessibility and reducing diagnostic barriers. The algorithm uses synchronized voltage gradient and photoplethysmographic signals. The goal was to match or exceed the performance of current diagnostic standards. The study aimed to validate the algorithm using clinical data. The algorithm was tested for sensitivity and specificity against gold-standard diagnostic methods. The purpose was to provide a front-line diagnostic tool with broader applicability.
Main Methods:
The study collected time-synchronized voltage gradient and photoplethysmographic signals from patients at rest. Signals were gathered for 230 seconds from subjects within seven days of diagnostic imaging. The data was paired with subject metadata and outcomes. A machine learning model was trained on a subset of 2,522 subjects. A cut point on the ROC curve was pre-specified for validation. An unseen test set of 965 subjects was used to evaluate the algorithm. Sensitivity and specificity were calculated at the selected cut point. The algorithm's performance was compared to existing diagnostic standards.
Main Results:
At the pre-specified cut point, the algorithm achieved a sensitivity of 0.73 (95% CI: 0.68-0.78). The specificity was 0.68 (95% CI: 0.62-0.74) at the same cut point. A different cut point yielded a negative predictive value of 0.99. At this point, sensitivity was 0.89 and specificity was 0.42. A cut point maximizing positive predictive value reached 0.12. This point had a sensitivity of 0.39 and specificity of 0.88. The algorithm's performance matched that of standard diagnostic tests. It offers advantages in non-invasiveness and patient comfort.
Conclusions:
The algorithm's performance is comparable to current diagnostic tests for coronary artery disease. It offers a non-invasive, radiation-free alternative to SPECT imaging. The system can be used in an office setting without patient stress. Multiple cut points allow customization of diagnostic thresholds. The negative predictive value matches that of coronary computed tomography angiography. The positive predictive value approaches that of myocardial perfusion imaging. The algorithm may serve as a front-line diagnostic tool. It provides advantages in accessibility and patient experience.
The algorithm uses synchronized voltage gradient and photoplethysmographic signals to detect coronary artery disease.
The test does not require radiation, expensive equipment, or induced patient stress.
The 230-second period was selected to capture sufficient physiological data at rest.
The ROC curve was used to determine optimal diagnostic thresholds for sensitivity and specificity.
At one cut point, the negative predictive value reached 0.99, matching coronary computed tomography angiography.
The authors propose that the algorithm may serve as a front-line diagnostic tool for coronary artery disease.