Machine-learning physiologic signal model achieved 90% sensitivity and 99% negative predictive value for ischemia detection in symptomatic patients

The model showed strong sensitivity in both epicardial coronary artery disease and suspected ischemia with nonobstructive coronary arteries – often referred to as INOCA

BETHESDA, Md.: CorVista Health, a leader in noninvasive cardiovascular diagnostics, today announced new data evaluating a physiologic signal–based machine-learning model for detecting cardiac ischemia. The data was presented at the American College of Cardiology’s Annual Scientific Session (ACC.26), taking place March 28–30, 2026 in New Orleans.

The study, titled “Noninvasive Ischemia Detection in Symptomatic Patients: A Physiologic Feature Machine-Learned Model,” evaluates a novel approach that uses physiologic signals and advanced machine learning to identify ischemia in patients presenting with symptoms suggestive of coronary disease.

Results showed strong diagnostic performance for detecting ischemia, including among patients whose disease can be difficult to identify using traditional imaging.

Key findings from the analysis include:

• Area under the curve (AUC): 0.86
• Sensitivity: 90%
• Specificity: 59%
• Negative predictive value: 99%

The model demonstrated consistent performance across patient subgroups including sex, age, and body mass index, and showed strong sensitivity in both epicardial coronary artery disease (eCAD) (91%) and suspected ischemia with nonobstructive coronary arteries (86%) – often referred to as INOCA. While eCAD is driven by obstructive plaque in the major coronary arteries, INOCA describes ischemia that occurs without visible large-vessel blockage, often reflecting dysfunction in the smaller vessels or coronary blood flow regulation.

“This research demonstrates the power of CorVista’s physiologic signal analysis combined with machine learning to help clinicians identify ischemia earlier and with greater confidence,” said Charles R. Bridges, MD, EVP and Chief Scientific Officer of CorVista Health. “A highly sensitive, noninvasive rule-out approach could help clinicians improve the diagnosis of coronary ischemia and guide more efficient downstream testing for symptomatic patients.”

Women with coronary ischemia often face a different diagnostic reality than men. Studies show that up to two-thirds of women referred for coronary angiography have no obstructive coronary artery disease, despite ongoing symptoms and risk, a ratio that is reversed in meni. As a result, women are more likely to undergo invasive testing that returns “normal” findings, while underlying conditions like INOCA often remain undetected. Yet, INOCA has similar prevalence in women to eCAD and INOCA is nearly twice as prevalent in women than in men, representing a huge unaddressed public health problem.

Currently, only Positron Emission Computed Tomography (PET-CT) can detect INOCA non-invasively, however PET-CT remains unavailable to >90% of patients with symptoms. Although invasive left heart catheterization (LHC) can detect INOCA definitively, the provocative maneuvers necessary to identify INOCA are not performed in the vast majority of LHC cases and screening for INOCA (SOC) remains outside the standard of care for most invasive cardiologists. Improving noninvasive identification of both INOCA and eCAD could help reduce unnecessary procedures and ensure women receive more accurate, timely diagnosis and care, resulting in decreased healthcare costs and reductions in mortality and morbidity.

“Clinicians need better ways to evaluate symptomatic patients with confidence; without defaulting to invasive, costly procedures that can still leave potentially life-threatening disease undetected,” said Adrian Lam, President and CEO of CorVista Health. “These findings underscore the potential for the CorVista System to help clinicians make faster, more accurate decisions while ensuring patients, especially those too often overlooked, receive earlier answers, fewer unnecessary procedures, and a clearer path to appropriate care.”

The findings highlight the potential for physiologic signal–based diagnostics to address a persistent gap in cardiovascular care – helping clinicians identify coronary ischemia of all types earlier and more confidently, while enabling patients to receive clearer answers and more appropriate next steps without unnecessary invasive testing.

About CorVista System®

The CorVista System is an FDA-cleared, non-invasive cardiovascular diagnostics platform designed to analyze cardiac and hemodynamic signals using machine-learned algorithms. The system synchronously collects physiological signals during a brief point-of-care test and applies advanced analytics to identify patterns associated with pulmonary hypertension (PH) and coronary artery disease (CAD). The CorVista System delivers actionable diagnostic insights without the use of radiation, contrast agents, injections, fasting, or exercise. The CorVista System is developed and manufactured by Analytics For Life, Inc. and licensed to CorVista Health, Inc.

About CorVista Health

CorVista Health is dedicated to transforming cardiovascular care through innovative diagnostics that shorten the path from symptoms to diagnosis. By enabling earlier detection of complex cardiovascular conditions, CorVista aims to empower clinicians with actionable insights and improve patient outcomes across diverse care settings.

For more information on CorVista Health, please visit: www.corvista.com

i Bairey Merz, C. N., Pepine, C. J., Walsh, M. N., Fleg, J. L., Camici, P. G., Chilian, W. M., ... & Wenger, N. (2017). Ischemia and no obstructive coronary artery disease (INOCA) developing evidence-based therapies and research agenda for the next decade. Circulation, 135(11), 1075-1092.

Fonte: Business Wire