An AI model trained on European pregnancies might miss the signals that matter in Indian ones
In New Delhi, India has taken a deliberate step toward confronting one of its most persistent maternal health burdens — the premature arrival of children who face a lifetime shaped by that early vulnerability. Through the GARBH-INi initiative, 12,000 women have offered their biological stories to science, allowing researchers to train artificial intelligence on the particular rhythms of Indian pregnancies. It is an act of collective trust: that data gathered today might spare future families the grief that preterm birth so often brings.
- India bears a disproportionate share of the world's preterm births, and the toll — neonatal death, developmental delay, chronic illness — has long outpaced the country's ability to predict or prevent it.
- The GARBH-INi study has quietly assembled one of the most comprehensive pregnancy data repositories in the world, with 1.6 million biospecimens and over a million ultrasound images now ready to train AI systems.
- Researchers have already produced working tools: AI models that date pregnancies more accurately for Indian women, microbiome signatures that flag high-risk cases, and genetic markers that could trigger early intervention.
- Technology transfer agreements with private firms signal the pivot from laboratory discovery to clinical deployment — AI ultrasound systems and microbiome-based therapies are now moving toward hospitals and clinics.
- Officials acknowledge the harder work lies ahead: translating promising models into routine medical practice across a vast and varied healthcare system.
India has formally launched its largest pregnancy study, enrolling 12,000 women under the GARBH-INi initiative to harness artificial intelligence and biological data in the fight against preterm birth. Presented this week at India Habitat Centre in New Delhi by Union Minister Dr. Jitendra Singh, the program addresses a condition that remains a leading cause of neonatal death and lifelong disability — one that falls with particular weight on India, which carries a disproportionate share of the global burden.
Preterm birth leaves immediate marks — respiratory failure, infection, organ fragility — and longer ones too: developmental delays, learning difficulties, chronic disease into adulthood. Because standard research has largely been conducted on non-Indian populations, officials determined that effective solutions would need to be built from Indian data, reflecting the genetic and nutritional realities of Indian pregnancies.
Over several years, GARBH-INi has assembled more than 1.6 million biospecimens and over one million ultrasound images from its enrolled women — forming the foundation for AI systems trained to detect which pregnancies are most at risk. The research has already yielded concrete results: more accurate pregnancy-dating models for Indian women, microbiome signatures linked to preterm risk, and early-stage diagnostic tools that could allow timely clinical intervention.
This week, the government formalized technology transfer agreements with three private companies. Sundyota Numandis Probioceuticals will develop microbiome-based therapies, while DOTO Health and Qure.ai Technologies will commercialize AI ultrasound and risk-prediction platforms for broader clinical use. A data-sharing platform, GARBH-INi-DRISHTI, has also been established to extend the study's findings to researchers across India and beyond.
Senior health official Dr. V.K. Paul noted that the real challenge now is deployment — moving these tools from research settings into the clinics and hospitals where they are needed most. For the 12,000 women whose pregnancies formed the raw material of this effort, their contribution was formally acknowledged: their trust in science now underpins India's hope of preventing thousands of preterm births in the years ahead.
India has launched what officials are calling the country's largest pregnancy study, enrolling 12,000 women in an effort to use artificial intelligence and biological data to predict and prevent preterm births—a condition that remains one of the leading causes of death and lifelong health problems in newborns. The initiative, known as GARBH-INi, was formally presented this week at India Habitat Centre in New Delhi by Union Minister Dr. Jitendra Singh, who oversees science and technology policy for the government.
Preterm birth is a significant public health burden in India. Children born too early face immediate risks of respiratory distress, infection, and organ failure. Those who survive often carry consequences into adulthood—developmental delays, learning disabilities, chronic lung disease. The problem is not unique to India, but India carries a disproportionate share of the global preterm birth burden, which is why officials decided the country needed solutions designed specifically for Indian populations rather than relying on research conducted elsewhere.
The GARBH-INi program represents a shift toward what researchers call a data-driven approach. Over several years, the study has collected biological samples—blood, tissue, and other specimens—from the 12,000 enrolled women, accumulating more than 1.6 million well-characterized biospecimens. Researchers have also gathered over one million ultrasound images from these pregnancies. This repository of material and imagery forms the foundation for training artificial intelligence systems to recognize patterns that predict which pregnancies are at highest risk of ending prematurely.
The research has already produced several concrete tools. Scientists have developed AI models that can date pregnancies more accurately for Indian women, accounting for genetic and nutritional variations that make standard dating methods less reliable. They have identified specific microbiome signatures—patterns in the bacteria living in the pregnant body—that appear to predict preterm birth risk. They are working on rapid diagnostic tests and genetic markers that could allow doctors to identify at-risk pregnancies early enough to intervene.
Beyond the research itself, the government has established what it calls the GARBH-INi-DRISHTI data-sharing platform, making the study's findings available to researchers across India and globally. This week, officials formalized several technology transfer agreements. One company, Sundyota Numandis Probioceuticals, will develop microbiome-based treatments based on the study's findings. Two other firms—DOTO Health and Qure.ai Technologies—have signed letters of intent to commercialize AI-enabled ultrasound systems and risk prediction platforms that could eventually reach hospitals and clinics across the country.
Dr. V.K. Paul, a senior health policy official, noted that the real work now lies ahead: taking the tools and models the study has produced and actually deploying them in clinical practice. He called for continued scientific collaboration and deeper analysis of the data already collected. The government has framed this effort as part of a larger mission to strengthen India's bioeconomy—which has grown from roughly 10 billion dollars in 2014 to about 195 billion dollars today—and to position India as a leader in preventive healthcare innovation.
For the 12,000 women who participated in the study, their contribution was formally acknowledged this week. Their pregnancies, their biological samples, their ultrasound images—all of it now forms the raw material from which India hopes to build tools that could eventually prevent thousands of preterm births and the suffering that follows. Whether these laboratory discoveries translate into widespread clinical benefit remains to be seen, but the infrastructure is now in place to find out.
Notable Quotes
India carries a significant share of the global burden of preterm births, making it essential to develop solutions suited to Indian conditions— Dr. Jitendra Singh, Union Minister of State for Science & Technology
The next phase should focus on effectively utilizing the tools, predictive models and platforms developed so far— Dr. V.K. Paul, Member, NITI Aayog
The Hearth Conversation Another angle on the story
Why does India need its own preterm birth research? Couldn't they just use studies from other countries?
Because the biology isn't universal. Genetics, nutrition, climate, infection patterns—they all vary. An AI model trained on European or American pregnancies might miss the signals that matter in Indian pregnancies. You need data from the population you're trying to help.
So they collected samples from 12,000 women. What exactly are they doing with all that biological material?
Training AI systems to recognize patterns. They have 1.6 million biospecimens and over a million ultrasound images. The algorithm learns: when these markers appear together, preterm birth is more likely. Then doctors can use that prediction to act—monitor more closely, give medication, plan earlier delivery if needed.
Is this actually going to help people, or is it just research that sits in a journal?
That's the real question. They've already signed deals with private companies to turn the findings into actual products—ultrasound systems that flag risk, probiotic treatments based on microbiome research. But Dr. Paul was clear: the tools exist now, but getting them into hospitals and clinics is the next hard part.
How many preterm births could this prevent?
They haven't said a number. But preterm birth is one of the leading causes of neonatal death in India. If even a fraction of pregnancies could be identified as high-risk early enough to intervene, the impact would be substantial.
Why frame this as nation-building? Why not just call it a health study?
Because in India's policy conversation right now, everything connects to the vision of a developed India by 2047. Healthier children today become more productive adults tomorrow. It's not wrong—it's just how the government talks about long-term investment.