Smartphone Based AI Outperforms Conventional ECG Methods in Detecting Hidden Heart Attacks During Clinical Trials

New AI ECG algorithms identify occlusive heart attacks with 84% accuracy, far surpassing traditional human diagnosis for patients without ST elevation.

By: AXL Media

Published: Mar 24, 2026, 5:39 AM EDT

Source: Information for this report was sourced from European Society of Cardiology

Revolutionizing the Detection of Silent Cardiac Occlusions

The traditional diagnostic pipeline for heart attacks is being challenged by artificial intelligence that can see what the human eye often misses. According to research presented at the European Society of Cardiology congress, an AI based ECG interpretation system has demonstrated superior accuracy in identifying occlusive myocardial infarctions (MI) compared to standard medical protocols. While clinicians rely heavily on ST elevation as a primary indicator of a blocked coronary artery, many patients suffer from life threatening occlusions without this specific electrical signal. The AI provides a critical second look, offering a way to bypass the uncertainty that often leads to dangerous delays in emergency cardiac intervention.

Bridging the Diagnostic Gap in Acute Coronary Syndrome

For patients presenting with chest pain but no ST elevation, the medical path forward is often slow and reliant on secondary testing. Doctor Federico Nani of Central Hospital Bolzano in Italy noted that recognizing these specific heart attacks quickly is a significant hurdle for frontline clinicians. Because these patients do not meet the classic criteria for an ST elevation myocardial infarction (STEMI), they frequently wait for hours while cardiac biomarkers like troponin are processed. The study sought to determine if an AI algorithm could analyze the initial ECG immediately upon arrival to optimize management and ensure that those with true blockages receive percutaneous coronary intervention without the usual wait times.

Analyzing Large Scale Patient Data with Precision

The prospective study involved a significant cohort of 1,490 patients, all of whom presented with symptoms of acute coronary syndrome but lacked the definitive ST elevation on their first ECG. With a mean age of 63 and a diverse gender split, the group represented a standard cross section of emergency cardiac patients. While doctors performed traditional troponin tests and angiographies, a CE certified AI-ECG algorithm simultaneously processed the same initial data via a smartphone based platform. This parallel testing allowed researchers to directly compare the speed and reliability of the machine against the standard ESC guidelines followed by human medical teams.