A routine heart test that takes less than two seconds to record is now being read by artificial intelligence well enough to catch disease that doctors themselves can miss. Two separate AI systems unveiled this year are turning the humble electrocardiogram (ECG), a test already performed on millions of patients annually, into a much more powerful screening tool for hidden heart failure and valve disease.
The most striking results come from a team at Imperial College London, presented at the European Society of Cardiology congress in Munich. Their AI model was trained to scan the same simple ECG trace a doctor already looks at, but to spot patterns invisible to the human eye. In a trial covering 67,000 patients in the United States, the tool correctly flagged up to 81% of heart failure cases and up to 90% of valve disease cases, even though a standard ECG was never designed to detect either condition.
That distinction matters. Confirming heart failure or valve disease normally requires an echocardiogram, an ultrasound scan of the heart that can take weeks or months to schedule in many health systems. An ECG, by contrast, is quick, cheap and already routine in emergency rooms, GP surgeries and pre-surgery checkups. If an algorithm can reliably flag which of those routine scans hide a deeper problem, patients who need an echocardiogram most urgently could be pushed to the front of the queue instead of waiting in line behind everyone else.
The Imperial College result is not an isolated case. In the United States, a separate AI system called EchoNext received FDA clearance earlier this year as the first artificial intelligence tool cleared to detect six distinct types of structural heart disease from a standard 12-lead ECG. Regulators approving a tool that reinterprets a decades-old test, rather than requiring new equipment, is itself part of a broader pattern: international regulators are increasingly racing to keep pace with software that finds new value in data institutions already collect.
Cardiologists caution that these tools are not a replacement for a doctor’s judgment. Both systems are designed to flag patients for further testing, not to issue a diagnosis on their own. But by working from a test already built into routine care, they avoid one of the biggest obstacles to catching heart disease early: cost and access. No new hardware is required in most clinics, which is part of a wider trend of artificial intelligence quietly working its way into everyday devices rather than arriving as a separate gadget.
Heart failure in particular is notoriously easy to miss in its early stages, since symptoms like fatigue or breathlessness are often blamed on age or being out of shape. Studies cited alongside the Imperial College trial suggest a meaningful share of patients only receive a diagnosis after an emergency hospital visit, by which point the condition is harder to manage. Tools that can flag risk from a scan already sitting in a patient’s file, without any extra test, could change that timeline. The push for smaller, more efficient hardware extends beyond software, too, echoing recent battery breakthroughs aimed at wearable medical devices that could one day run continuous heart monitoring outside a clinic altogether.
For now, both tools remain in the early stages of clinical rollout, and neither is available for patients to request directly. Researchers say the next step is testing the models in more diverse hospital settings outside the original trials, to confirm the accuracy holds up across different populations and equipment. If it does, a test that has existed largely unchanged for a century may become one of the more unlikely frontiers in health and medicine, alongside broader advances in science and technology and high-tech digital tools reshaping everyday care.
Questions frequently asked
What exactly does the new AI heart tool do?
It analyzes the electrical trace from a routine ECG, a test that normally checks heart rhythm, and looks for subtle patterns linked to heart failure or valve disease that are not part of what the test was originally designed to measure.
Does this replace the need for an echocardiogram?
No. The AI tools are designed to flag which patients should be prioritized for an echocardiogram or further testing, not to replace that diagnostic step. A positive flag still requires confirmation through standard clinical evaluation.
Is this technology available to patients now?
One system, EchoNext, has received FDA clearance in the United States for six types of structural heart disease detection. The Imperial College London model was presented at a medical congress and is still moving through further validation before wider clinical use.
Sources
The Guardian,
Dataconomy,
Medical Daily
This article was written with the help of artificial intelligence. Editorial policy
