AI Turns Routine 10-Second ECGs Into Predictive Scans for Heart Disease and Diabetes

A routine electrocardiogram (ECG) takes just ten seconds and costs pennies to administer. For decades, it has been a standard but relatively blunt instrument used primarily to check heart rate and rhythm. Now, a spinout from Imperial College London has secured £1.4 million in funding to deploy artificial intelligence that transforms this basic test into a comprehensive diagnostic radar.[1][2]

The newly launched company, Cardiovolt.ai, aims to commercialize deep learning models that read the subtle electrical signals hidden within an ECG trace. These models can detect underlying heart failure, valve disease, and even non-cardiovascular conditions like diabetes and chronic kidney disease long before clinical symptoms appear.[1][3]

"We looked at the ECGs to see if we could do things that are superhuman," said Dr. Arunashis Sau, Chief Scientific Officer of Cardiovolt.ai and a cardiology registrar at Imperial College Healthcare NHS Trust. He emphasized that the goal is not to replicate what clinicians already do, but to extract insights that no human cardiologist, regardless of their expertise, can see with the naked eye.[1][2]

The AI achieves this by identifying "digital biomarkers"—minute waveform patterns embedded in the heart's electrical trace that correlate with broader systemic disease. To train the neural networks to recognize these invisible signatures, researchers required data at a massive scale.[2][3]

The foundational models were trained on a databank of over 1.6 million ECGs provided by a research group in Brazil, each meticulously linked to the patient's longitudinal medical history. This was supplemented by several million additional ECG records from the United States, providing the AI with a globally diverse dataset to learn the higher-dimensional relationships across millions of heartbeats.[1][2]

The clinical validation results have been striking. In international testing, the AI models achieved a diagnostic accuracy of between 83 and 93 percent for hidden heart diseases, such as low ejection fraction and undiagnosed heart failure.[1][3]

Even more surprisingly, the models demonstrated a 70 to 80 percent accuracy rate in flagging non-cardiovascular conditions, including diabetes and kidney disease, purely from the electrical activity of the heart. The models can also estimate a patient's short- and long-term mortality risk, providing a holistic health snapshot from a single ten-second recording.[1][2]

The technology builds on years of foundational research supported by the British Heart Foundation and the National Institute for Health and Care Research (NIHR). Previous iterations of the team's work, such as the AIRE model published in late 2024, successfully predicted the risk of early death from ECGs, while a 2025 tool called AIRE-CHB identified patients at high risk for a fatal condition known as complete heart block with 89 percent accuracy.[5][6]

By spinning out into a dedicated commercial entity, the Imperial College team hopes to accelerate the technology's integration into actual hospital workflows. Professor Fu Siong Ng, Cardiovolt.ai's Chief Medical Officer, noted that the spinout route was chosen because it offers the fastest path to navigating regulatory approvals and deploying the software directly into electronic health record systems.[2][3]

The immediate clinical focus will be on proactive diagnosis. Currently, patients often undergo expensive and time-consuming echocardiograms or MRI scans only after they present with severe symptoms. Under the new AI-enhanced pathway, a routine ECG conducted during a standard checkup could instantly flag high-risk patients, prompting immediate follow-up imaging and early intervention.[2][4]

Health technology analysts point out that this approach could fundamentally shift population health strategies. By converting an inexpensive, ubiquitous test into a powerful screening tool, healthcare systems can prioritize patients most at risk without requiring massive investments in new diagnostic hardware.[3][4]

This hardware-agnostic approach holds particular promise for global health equity. In under-resourced clinical settings or developing nations where access to advanced imaging equipment is scarce, a standard digital ECG machine paired with cloud-based AI analysis could deliver expert-level diagnostic triage to millions.[3][5]

The company is currently pursuing partnerships with health systems and medical device manufacturers to embed the algorithms directly into hospital networks. If deployed at scale, the technology promises to turn one of medicine's oldest and most common tests into one of its most informative, saving lives through the power of early detection.[1][3]