CHENNAI: Dr Ziad Obermeyer, an emergency medicine specialist and associate professor at the School of Public Health at University of California, Berkeley, is leading a clinical trial in Tamil Nadu that is testing whether artificial intelligence can help detect so-called silent heart attacks, a condition that often goes unnoticed until it causes serious long-term damage.
Silent heart attacks, known medically as unrecognised myocardial infarctions, occur without the classic warning signs such as severe chest pain. Many people who experience them do not seek medical care, allowing damage to the heart muscle to progress quietly over time. This significantly raises the risk of heart failure, irregular heart rhythms and future cardiac events, making early detection a major challenge for doctors, particularly in resource-constrained settings.
The trial in Tamil Nadu is examining whether AI can identify subtle patterns in electrocardiogram readings that point to a past heart attack, even when a patient has never been diagnosed. The research team is using simple, low-cost ECG devices, similar to portable machines increasingly used in clinics and community health programmes, and pairing them with machine-learning models trained to spot signals that are difficult for the human eye to detect. The goal is not to replace standard diagnostic tests, but to use AI as an initial screening tool that can flag high-risk patients for further evaluation.
The trials are being conducted within the State's public healthcare system, rather than as an open screening exercise across the general population. The study is taking place in selected government hospitals and affiliated health facilities in the state. Participants are typically patients who are already visiting these hospitals for routine care or other medical reasons, such as people coming to outpatient departments or emergency units, rather than healthy volunteers recruited from the community at large. This allows researchers to test the AI tool in real-world clinical settings where ECGs are commonly used.
As part of the trial, patients undergo standard ECG tests using low-cost or portable devices, and the AI system analyses these readings to look for signs of a past, previously undetected heart attack. The results generated by the AI are not used on their own to make clinical decisions; instead, they are compared against existing medical records, clinician assessments and, where necessary, further diagnostic tests. This helps researchers evaluate how accurately the AI can flag silent heart attacks in routine care environments.
Researchers say this hospital-based approach is deliberate. Testing the tool in government hospitals ensures the technology is evaluated among a broad cross-section of the population, including lower-income patients who are often at higher risk of undiagnosed heart disease. It also reflects how the system would likely be used if scaled up, as part of primary or secondary healthcare rather than through mass screening camps.
The trial is being carried out with ethical approvals and informed consent from participants, and patient data are anonymised. The longer-term aim is to assess whether such AI tools can be safely and effectively integrated into everyday clinical workflows in public hospitals, before considering wider deployment at the community level.
Researchers involved in the study say the approach could be especially valuable in primary healthcare settings, where access to advanced imaging and specialist cardiology services is limited. By identifying patients who may already have heart damage, doctors can begin preventive treatment earlier, including medication to control blood pressure and cholesterol, closer monitoring and lifestyle interventions that reduce the risk of future heart attacks.
Tamil Nadu was chosen as the trial site because it offers a large and diverse population, a relatively strong public health system and an environment well suited for large-scale clinical research. Conducting the trial in India also helps ensure that AI models are trained and tested on data from non-Western populations, addressing a common criticism that many health algorithms are built using datasets that do not reflect global diversity.
Beyond detecting silent heart attacks, Dr Obermeyer’s team is also exploring whether similar AI-based tools could eventually help identify other dangerous conditions, such as ongoing heart attacks or blood clots, in real time in emergency settings. Such tools could support faster decision-making by clinicians, particularly in busy hospitals and clinics.
If successful, the Tamil Nadu trial could demonstrate how artificial intelligence, combined with basic medical technology, can strengthen early detection of heart disease in low- and middle-income regions. Public health experts say this kind of approach has the potential to reduce preventable deaths and long-term disability from cardiovascular disease, which remains one of the leading causes of mortality in India and globally.