AI Tool Reads ECGs in Under Two Seconds, Flags Heart Disease

What Happened

Researchers have developed an artificial intelligence tool capable of detecting heart disease from a routine electrocardiogram in less than two seconds, according to a report published Sunday. The system, described as performing at a level above that of trained clinicians in certain diagnostic tasks, was trained on millions of ECG recordings and is designed to identify high-risk patients who may need to be fast-tracked for further treatment.

How the Technology Works

The tool analyzes standard ECG data, the kind collected during routine clinical visits, and processes the signal to flag patterns associated with heart disease. Developers say the system completes its analysis in under two seconds per reading. The training dataset comprised millions of ECGs, giving the model exposure to a broad range of cardiac presentations across large patient populations.

The Guardian reported that the tool performs at what developers characterize as a superhuman level for specific detection tasks, meaning it identified conditions in ECG data at a rate that exceeded the performance of human clinicians in controlled evaluations. The report did not specify the precise conditions detected or the clinical settings in which the comparisons were made.

Background

ECG-based cardiac screening has long been a standard component of clinical care, but its diagnostic value depends heavily on the availability and expertise of trained readers. In many health systems, specialist review creates bottlenecks that delay diagnosis and treatment, particularly for patients in primary care or emergency settings where cardiology expertise may not be immediately available.

AI-assisted ECG analysis has been an active area of research and commercial development for several years. Companies including AliveCor and Cardiologs, as well as academic medical centers, have produced systems that assist clinicians in reading ECG data. The technology reported Sunday represents a continuation of that research trajectory, with developers emphasizing speed and accuracy as distinguishing features.

Cardiovascular disease remains the leading cause of death globally, according to the World Health Organization, which reported in 2023 that an estimated 17.9 million people die from cardiovascular conditions each year. Earlier identification of high-risk patients is a recognized clinical priority.

What It Means in Practice

The developers indicate the tool is intended to support clinical workflows by rapidly sorting patients based on cardiac risk, enabling clinicians to prioritize those most in need of specialist evaluation or intervention. The system would function as a triage aid rather than a replacement for physician review, operating on data already collected through existing diagnostic equipment.

If integrated into standard care pathways, a sub-two-second analysis window could allow the tool to process large volumes of ECGs quickly, including in settings such as emergency departments or remote clinics where timely review is constrained by staffing. The Guardian report noted the technology could fast-track high-risk patients for treatment, though specific deployment timelines and healthcare system partnerships were not detailed in available reporting.

The use of large training datasets is consistent with approaches taken by other high-performing medical AI systems. Researchers in adjacent fields, including retinal imaging and radiology, have demonstrated that model performance scales with both dataset size and diversity, though generalizability across different patient populations and recording equipment remains an ongoing area of study.

What Comes Next

Further peer-reviewed publication of the clinical validation data, along with any regulatory submissions in the United Kingdom, European Union, or United States, would be required before the tool could be used in routine patient care in those jurisdictions.