Artificial Intelligence for Cardiac Ischemia Detection: A Review of Biosensor, Biomarker, and Imaging Applications

Abstract

Ischemic heart disease is the leading cause of global mortality, yet accurate and timely detection remains constrained by incomplete signal capture, noise, operator dependence, and imperfect reference standards. Artificial intelligence (AI) may enhance cardiovascular sensing by extracting latent diagnostic information from routinely acquired signals. This narrative review examines AI-enabled approaches to ischemia detection across three sensing modalities: the electrocardiogram as a bioelectric sensor, cardiac troponin as a biochemical marker, and cardiac imaging technologies (echocardiography, computed tomography, magnetic resonance, and nuclear imaging) as structural and functional sensors. For each modality, we evaluate the sensing pipeline, the reference standards, the endpoint actually measured (from acute coronary occlusion to functional ischemia), and reported performance metrics, separating diagnostic accuracy from effects on clinical workflow, patient outcomes, and cost. We also discuss emerging directions, including foundation models and multimodal fusion, together with translational challenges such as limited prospective validation, regulatory hurdles, and uncertain cost-effectiveness. Most evidence comes from retrospective, frequently single-center studies of diagnostic performance; few studies have measured effects on workflow, and fewer still on patient outcomes or cost. Routine clinical adoption will require prospective multicenter trials, standardized performance benchmarks, and evidence of clinical and economic benefit.