An independent
healthcare AI startup.

We're building software that takes an ECG recording and estimates the person's blood glucose level, using a signal that hospital monitors, ECG machines and wearables already record, instead of a finger-prick or a sensor under the skin.

Checking blood glucose usually means a finger-prick or a sensor inserted under the skin. That cost and discomfort limit how often people test, and many people with diabetes remain undiagnosed. Daibeats is developing technology that estimates glucose from the electrocardiogram (ECG), a signal already recorded by hospital monitors, clinical ECG machines, and a growing number of wearables.

Our approach combines heart rate variability and ECG waveform features with machine learning models trained with safety-aware objectives, so that errors in clinically dangerous glucose ranges are weighted more heavily. We are building software rather than new hardware.

Our first users will be research and clinical partners who need to analyse ECG and glucose data together. In the planned product, Claude, Anthropic's AI model, will turn each estimate and the ECG features behind it into a short plain-language report for the researcher or clinician. Over time, we aim to develop this into a practical healthcare product, following the clinical evidence and regulatory steps that requires. How the technology works →

Daibeats is an independent startup, not a university lab. The technology originates from research in ECG signal intelligence by the founder at Tezpur University, carried out with co-author S. I. Singh. We cite that academic work openly, and the people credited on it, rather than presenting it as company output.

Earlier research outputs that appear on this site, including papers dated before 2026, are the research foundation the company builds on. They are not products or commercial results of Daibeats.