The short version
- The AI tool demonstrated high accuracy in identifying heart failure and valve disease from standard ECG data in a large US patient trial.
- Researchers propose using the system to fast-track urgent cases for echocardiograms, reducing delays that currently leave patients waiting months for diagnosis.
- While not a standalone diagnostic method, the technology aims to flag high-risk individuals who might otherwise remain undiagnosed until symptoms become severe.
A newly developed artificial intelligence system has shown the ability to detect signs of serious heart conditions from standard electrocardiogram readings in under two seconds. The technology, which was presented at the European Society of Cardiology annual congress in Munich, represents a significant shift in how routine cardiac data might be utilized. By extracting subtle patterns that are typically invisible to human clinicians, the tool could help identify patients with heart failure or valve disease much earlier than current standard practices allow.
Traditional electrocardiograms have served as a cornerstone of cardiac care for over a century, effectively recording the heart’s electrical activity to diagnose attacks and rhythm abnormalities. However, these tests are not designed to detect structural issues like heart failure or valve dysfunction. Confirming such conditions usually requires an echocardiogram, an ultrasound scan that provides detailed images of the heart’s structure and function. Access to these scans is often limited, with patients frequently facing waiting periods of several months before they can be examined.
The new AI model addresses this bottleneck by analyzing ECG data to flag potential structural problems. In a trial involving 67,000 patients in the United States, the system identified up to 81 percent of individuals with heart failure and up to 90 percent of those with heart valve disease. These figures suggest that the tool could serve as a powerful screening mechanism, highlighting those who are most likely to have an abnormality so they can be prioritized for further testing.
Experts emphasize that the technology is not intended to replace definitive diagnostic procedures. Instead, it functions as a triage aid. When the AI indicates a high probability of heart disease, patients can be sent rapidly for echocardiograms rather than remaining on standard waiting lists. This acceleration in care could enable earlier initiation of lifesaving medications, which is critical for managing conditions that often progress silently until they become dangerous.
Dr. Sonya Babu-Narayan, a consultant cardiologist and clinical director at the British Heart Foundation, noted that while the tool will not detect every case of heart disease, it offers a practical solution for fast-tracking high-risk patients. The foundation funded the trial, recognizing the potential for earlier intervention to improve patient outcomes. The ability to process ECG results almost instantly could transform routine checks into more comprehensive screening opportunities.
Professor Fu Siong Ng from Imperial College London highlighted the logistical benefits of this approach. Current referral systems often result in significant delays between a doctor’s suspicion and the actual imaging scan. By identifying at-risk patients through AI analysis, healthcare providers can allocate scarce ultrasound resources more efficiently. This prioritization ensures that those with the most urgent needs receive attention sooner, potentially reducing the burden on overstretched cardiac departments.
Beyond targeted referrals, researchers see potential for opportunistic screening. The AI model could be applied to all ECGs performed in hospital settings, even when heart failure or valve disease is not the primary concern. This broad application might uncover undiagnosed conditions in patients who are undergoing tests for unrelated reasons. Such incidental findings could lead to earlier treatment and better long-term health outcomes for individuals who were previously unaware of their cardiac status.
The development is part of a broader trend in medical AI, with other researchers presenting similar advancements at the same conference. For instance, studies from Tokyo institutions demonstrated that AI analysis of short facial videos could detect undiagnosed high blood pressure and type 2 diabetes. These parallel innovations suggest a growing capacity for machine learning to extract diagnostic information from diverse, non-invasive data sources.
Looking ahead, the team behind the ECG tool aims to develop handheld devices that integrate this AI capability for use by healthcare professionals in various settings. Dr. Ahmed El-Medany, who led the analysis at Imperial College London, described the current system as superhuman in its speed and pattern recognition. The next phase involves making this technology accessible and practical for widespread clinical adoption.
While the results are promising, the tool remains a screening aid rather than a definitive diagnostic instrument. It cannot rule out or confirm heart disease on its own. Clinical validation and integration into existing healthcare workflows will be necessary to realize its full potential. Nevertheless, the ability to rapidly flag high-risk patients from routine ECGs offers a compelling path toward earlier diagnosis and more timely treatment for common forms of heart disease.
Sources behind this briefing
Go to the original reporting
- The Guardian US↗‘Superhuman’ AI tool spots heart disease in less than 2 seconds
