In the vast landscape of rural India, where a government hospital can be hours away and the nearest doctor may be a name on a board rather than a face in the clinic, one figure has remained constant for over two decades: the ASHA worker.
Accredited Social Health Activists (ASHAs) are the women who walk village lanes at dawn to check on pregnant mothers, carry registers of newborns and track vaccinations, and are often the first point of contact between a rural family and the formal healthcare system.
There are approximately one million of them across the country, and for most of the communities they serve, they are not merely health workers. They are the health system itself.
Now, a collaboration between the George Institute for Global Health and the University of Oxford is asking a question that sounds simple but represents a genuine departure from how health technology is usually built: what if the people who use a tool were also the people who design it?
Building from the ground up
The project centres on SMARThealth Pregnancy 2, an AI-powered mobile application currently being developed to help ASHA workers identify and manage high-risk pregnancies in rural communities.
When complete, the tool will support frontline workers in screening for complications, flagging women who need urgent referral, and following up on cases that might otherwise slip through the gaps of an overstretched system.
For millions of women in remote communities, ASHA workers are not just health workers but the health system itself. Photograph: (MIT Technology Review)
What distinguishes it from similar digital health initiatives is not the technology itself but the process behind it. Rather than developing the app in a research institution and rolling it out to workers as a finished product, the team has taken a bottom-up approach in which ASHAs are co-designers.
The workers are being involved in identifying what information they actually need, how it should be presented, what language and interface choices make sense given how they work, and where the real friction points in their daily practice lie.
This is a meaningful distinction. Community health workers in India have historically been given tools designed by people who have never done their jobs, and the results have often reflected that gap. Apps go unused, and registers are duplicated.
Digital systems sit alongside paper ones because the digital version does not quite map onto the reality of the work. Designing with end users rather than for them is a principle that has transformed product development in the technology sector, but it has been slow to arrive in global health.
Why maternal health in rural India needs this
The need is not difficult to establish. India’s maternal mortality ratio has improved substantially over the past two decades, falling from 254 per 100,000 live births in 2004 to 97 per 100,000 in 2018-20, according to government data. Still, the numbers continue to mask deep geographic inequity.
Rural and tribal communities carry a disproportionate share of the burden, and the reasons are familiar: distance from facilities, shortage of skilled birth attendants, low rates of antenatal care, and the kinds of complications — hypertension, anaemia, gestational diabetes — that are manageable when caught early and catastrophic when not.
A new AI pregnancy tracking tool from the George Institute and Oxford is changing the terms of that relationship — by putting ASHAs in the design room, not just the field. Photograph: (The Hindu)
ASHA workers are already the primary mechanism through which the government reaches these communities. They are trained to conduct home visits, counsel families on nutrition and hygiene, accompany women to facilities, and report complications up the chain.
The question the project seeks to answer is whether an AI tool, thoughtfully designed and in genuine partnership with the workers themselves, can extend what ASHAs can do without overwhelming them with yet another layer of administrative demands.
What co-design actually looks like
The co-design process involves iterative rounds of engagement with ASHA workers, in which the researchers observe practice in the field, present interface prototypes, gather structured feedback, and revise accordingly.
The workers are contributing knowledge about how pregnancies are discussed and understood in their communities, about the literacy levels and digital comfort of the women they visit, and about the practical constraints of working in areas with poor connectivity and unpredictable light.
The intention is to produce a tool shaped by the texture of real work rather than an idealised version of it — one that accounts for the fact that an ASHA conducting a visit may be standing in a small room with three generations of a family, holding a phone in uncertain light, and making a judgment call that will determine whether a woman gets referred to a facility that is an hour’s journey away.
The larger argument this project is making
What the George Institute and Oxford are demonstrating with this project goes beyond maternal health. India’s rural healthcare system has long relied on the labour and local knowledge of frontline women workers without adequately recognising or resourcing that contribution. The introduction of AI into that system carries both promise and risk.
The promise is that it could amplify the capacity of workers who are already stretched thin, giving them better information at the point of care. The risk is that poorly designed tools add burden, erode trust, or in the worst cases, substitute for the human judgment and relationship-building that make ASHAs effective in the first place.
Taking the co-design route does not eliminate those risks, but it significantly reduces them. A tool that ASHAs have helped shape from the ground up is more likely to fit the way they work, more likely to be used consistently, and more likely to surface the right information at the right moment.
