A team of cardiologists, endocrinologists, and AI researchers from a top-tier Indian medical institution has introduced DiaCardia an artificial intelligence model capable of detecting prediabetes and early-stage type 2 diabetes risk using only a standard 12-lead electrocardiogram (ECG), without requiring any blood draw. The breakthrough, published in early January 2026, addresses one of the biggest barriers to widespread prediabetes screening in India: the need for fasting blood tests, which are inconvenient, costly, and often skipped in routine check-ups or mass screening programmes.

Prediabetes affects an estimated 100–150 million Indians and is a major precursor to type 2 diabetes, cardiovascular disease, and other complications. Current screening relies on fasting plasma glucose, HbA1c, or oral glucose tolerance tests methods that require lab access, patient preparation, and time. DiaCardia changes this paradigm by training deep learning models on a massive dataset of more than 1.2 million ECGs linked to simultaneous lab-confirmed glucose values from diverse Indian populations.

The AI analyses subtle ECG features such as QRS morphology, T-wave abnormalities, heart rate variability, and interval changes that reflect early metabolic and autonomic dysfunction before overt symptoms or blood marker changes become apparent. After rigorous cross-validation and external testing on independent cohorts, DiaCardia demonstrated 87–92% sensitivity and 85–89% specificity for detecting prediabetes (IFG/IGT range), outperforming many traditional risk scores and matching or exceeding HbA1c-based screening in certain subgroups.

The model is designed to be lightweight and deployable on standard ECG machines or even smartphones equipped with portable ECG attachments. It generates an instant risk score and probability output along with explainable visual heatmaps highlighting the ECG segments most influential in the prediction. Importantly, the system does not diagnose diabetes outright but flags high-risk individuals for confirmatory blood testing and immediate lifestyle counselling making it a powerful screening and triage tool.

India has one of the highest burdens of prediabetes globally, yet screening remains opportunistic and low due to logistical barriers. DiaCardia turns a routine, widely available ECG into a non-invasive, zero-cost metabolic risk indicator that can be performed in seconds during any clinical encounter.

The model has been piloted in several urban and rural primary care centres in collaboration with state health departments. Early feedback shows high acceptability among physicians and patients, with the tool reducing the number of missed prediabetes cases and enabling earlier dietary, exercise, and pharmacological interventions.

Researchers are now planning larger multi-centre validation trials across different regions of India to confirm performance in varied ethnic, age, and comorbidity profiles. Regulatory clearance for clinical use as a Class B/C medical device is underway, with a target for commercial rollout in 2027.

The development is timely given India’s massive ECG infrastructure millions of ECGs are performed annually in government and private settings and the urgent need to curb the diabetes epidemic through affordable, scalable early detection.