IIT Madras Builds an Indian Formula for Foetal Age
IIT Madras says its AI model Garbhini-GA2, trained on Indian pregnancy data, could reach ultrasound scanners by 2027-28.

Clinicians in India estimate the age of a foetus from ultrasound scans using a formula built on Western populations. IIT Madras has built an Indian one, and says it could be available in scanning systems by 2027-28.
Prof Balaraman Ravindran, head of the Wadhwani School of Data Science and AI, told DT Next the model is meant to fix the limits of applying Western-derived formulae to Indian pregnancies. In his words, "We've developed an Indian babies' formula."
What exists today
The model is called Garbhini-GA2. It was developed and externally validated on data from Indian pregnancies, including validation with Christian Medical College, Vellore. Published research shows it cuts errors in gestational-age estimation compared with the widely used Hadlock formula. Work with CMC Vellore and Faridabad has produced papers already.
What is still hard
Ravindran was candid that the technology is ready but deployment is not. Open questions:
- Whether it can go into scanning machines at scale, beyond the pilot.
- Time. Ultrasound slots in busy government hospitals are around 90 seconds, and extra steps are stressful for staff.
- Compatibility with existing ultrasound equipment, and further validation, including possible work with Army hospitals.
The wider push
The foetal-age project sits inside a broader expansion. The school plans to strengthen AI for Science over the next two to three years, and is building an AI for Education initiative backed by the Union government, with tools for teachers in Indian languages, some designed to run on mobile devices with limited cloud dependence.
My read: this is what useful Indian AI looks like. A narrow model, local data, published validation, and an honest admission that workflow is the bottleneck. The model was the easy part.
Source: DT Next - Indigenous tech to determine foetal age in 2027-28, says IIT-Madras