As AI reaches rural India, health workers are putting it to the test
From screening for tuberculosis to monitoring newborns, AI is reaching rural India amid healthcare workforce shortages. But scaling it means fitting the technology into already stretched workers’ routines.
ASHA worker Tai Api records a newborn baby’s measurements using the AI-enabled Shishu Maapan app in Dadra and Nagar Haveli district, India, in 2024. (Photo: Wadhwani AI)
New: You can now listen to articles.
This audio is generated by an AI tool.
Read a summary of this article on FAST.
Get bite-sized news via a new
cards interface. Give it a try.
Click here to return to FAST
Tap here to return to FAST
FAST
BENGALURU: For years, Panthula Uha, a community healthcare worker in Chagallu village in the southern Indian state of Andhra Pradesh, knew what to ask to identify people who might have tuberculosis (TB). But this method of screening could miss villagers who showed no symptoms, she said.
That changed in March, when Uha and her team of Accredited Social Health Activists (ASHAs) – government-employed frontline health workers whose acronym also means ‘hope’ in Hindi – began going house to house, asking people to do something unusual: cough into a smartphone for a few seconds.
Their phones were equipped with an artificial intelligence-enabled app called Swaasa to detect patterns linked with respiratory diseases. Uha said the tool identified asymptomatic people for further testing, with some subsequently diagnosed with TB and put on treatment.
For T Gowramma, a 73-year-old patient at Chagallu village, this experience was less daunting.
“Whenever I had a cough, I used to worry that it was something serious,” she said. “The hospital staff came to our home and tested me using just my cough on the Swaasa app. It was very easy.”
Guess Word
Crack the word, one row at a time
Buzzword
Create words using the given letters
Mini Sudoku
Tiny puzzle, mighty brain teaser
Mini Crossword
Small grid, big challenge
Word Search
Spot as many words as you can
Hundreds of kilometres west, health workers are using an AI app called Shishu Maapan. It estimates a newborn’s weight and other measurements from a short video taken on a smartphone, replacing bulky equipment traditionally carried from home to home.
“At first, we wondered how weight could be captured on a mobile,” said Jyotsna Patel, an ASHA worker who has worked in Kachigam village in Daman district for 14 years.
After training and practice, she said, workers were able to use the app routinely and track whether babies were gaining or losing weight.
What connects these two instances in rural India is not the technology itself, but the healthcare workers using it.
Across the country, doctors, radiologists and community health workers in India’s public health system are already “so loaded with tasks”, said Sagar Sen, senior vice-president at Qure.ai, that adding another layer of work can defeat the purpose of the technology. The Mumbai-based firm builds AI tools to analyse medical scans for diseases like TB and lung cancer.
“The challenge is to build AI around the way they already work, so that it can take on parts of the workload while leaving clinical judgement with the healthcare worker,” he said.
In large parts of the country, where the first point of care for millions of people may be a community health worker or a doctor without a specialist down the corridor, AI is increasingly being built around the people already standing at that front line, experts said.
AI MEETS THE FRONTLINE IN RURAL INDIA
India has about 1.02 million ASHA workers, forming the world’s largest community health worker programme. Each is typically expected to serve about 1,000 people, with the programme spanning some 586,000 rural villages, according to government data.
India’s National Health Mission describes ASHA workers as the “first port of call” for health needs, particularly for women, children and others who struggle to access health services.
Their responsibilities already span maternal and child health, immunisation and TB screening, while newer duties such as household enumeration, maintaining health registers and data validation have added to their workload.
One of those responsibilities is monitoring newborns during home visits. In the first weeks of a baby’s life, an ASHA worker makes several such visits.
Workers can now use the Shishu Maapan app, which analyses a roughly 30-second smartphone video to estimate measurements such as weight, height and head circumference of babies up to 42 days old.
The process was initially unfamiliar to mothers of newborns, who expected their babies to be weighed on a scale as before, said Patel.
But some soon embraced the smartphone-based method.
“Now they lay the sheet down themselves and want to see if their baby has gained weight,” Patel said.
The results are also recorded digitally and can be synced with state systems rather than written into registers, said Dr Aparna Chaudhary, national lead for health at Wadhwani AI, which developed Shishu Maapan.
The measurements matter because low birth weight remains a significant problem in India.
About 18 per cent of babies are born weighing less than 2.5kg, while prematurity and low birth weight together account for 48 per cent of neonatal deaths, according to a Feb 2025 UNICEF India nutrition report.
“If a baby’s weight is low, we counsel the mother during the first visit itself,” Patel said, including advising mothers on exclusive breastfeeding.
Shishu Maapan’s AI model was trained on data from more than 30,000 infants to estimate a newborn’s measurements from a short video, Chaudhary said.
The AI app was built for ASHA workers using low-end Android phones and designed to work offline in areas with poor connectivity and tolerate imperfect lighting and movement, she added.
In another example, in Goa, Dr Govind Desai recalled how Qure.ai’s software flagged a small lung nodule within minutes on an X-ray taken for a patient’s shoulder injury.
With specialist radiologists scarce at smaller government health centres, X-rays previously sent to larger hospitals could take seven to 10 days to be reviewed, he said.
MAKING AI WORK FOR HEALTH WORKERS
For AI developers in India’s public health system, the challenge is not just getting the technology right, but ensuring it does not add to already stretched frontline workers’ burdens, experts said.
For Shishu Maapan, it had to be designed in a way that was simple enough to fit into an ASHA worker’s existing routine, rather than becoming another task added to an already crowded day, said Chaudhary.
A supervisor accompanied the workers into the field and showed them how to position the baby and record the video correctly. After practising, she said, they were able to use the app routinely and track whether a baby’s weight had increased or decreased.
Shishu Maapan has since been integrated into the local administration’s existing ASHA platform and handed over to the government to operate, Chaudhary said.
“For some ASHA workers, using the (Swaasa) app could be difficult,” said Uha, who leads a team of ASHAs, pointing to differences in age, education and familiarity with digital tools.
Uha said using the Swaasa app to screen people for TB initially felt like “a bit of a work”.
In the initial days, the app faced connectivity issues, while some people coughed directly in front of health workers, raising concerns about infection even though the workers were given masks, she said.
Dr Nammi Vasundhara, the East Godavari district TB and leprosy officer who oversaw the Swaasa pilot, told CNA that some workers were also uncomfortable asking people suspected of carrying an infectious disease to cough near their personal phones.
A larger government rollout could require dedicated devices for the tests, she said.
In Goa, doctors using Qure.ai are trained to interpret X-rays, understand what an AI flag means, follow the appropriate procedure and know where to refer the patient.
Qure.ai’s Sen said that AI should sit within the way healthcare workers already operate rather than create a parallel layer of work.
It may also help determine which AI experiments survive beyond a pilot, experts said.
FROM PILOT TO PRACTICE
Not every tool that works in a pilot makes it into routine government healthcare in India.
Salcit Technologies – developer of the Swaasa app – has tested it with public health workers and through non-profit programmes, but its chief executive Venkat Yechuri said the company has yet to secure a statewide or nationwide government rollout.
“We’ve had conversations with every state government that you can think of,” Yechuri said, “but we don’t know what exactly it takes to take it from a pilot to something else.”
Programmes that have gone further suggest scaling AI requires integrating into existing government systems, with clear pathways for workers and patients to act on its findings, experts said.
For instance, Qure.ai’s role in Goa’s government hospitals extends beyond flagging suspicious X-rays, Sen said.
A patient navigator – who helps coordinate a patient’s follow-up care – tracks AI alerts and guides flagged people through specialist review and further tests, including CT scans, he added.
The chest physician Desai said this has cut a process that previously took three to four weeks to around three to four days.
The coordination is particularly important when patients may have to travel around 40km for specialist care, Sen said.
He added that the firm had learnt from earlier deployments that an AI algorithm that generates more findings without a health system capable of dealing with them can simply create another bottleneck.
Similar experiments are underway elsewhere in Asia, where specialist care remains uneven.
In the Philippines, Qure.ai’s technology is used in mobile X-ray vans under the ACCESS TB Project to screen remote communities, with 8,700 TB cases diagnosed and notified through the programme, according to the company.
The company’s software was also used in Vietnam to screen over 96,000 people for TB.
Beyond Southeast Asia, Qure.ai has also deployed the technology across Africa.
Sen suggests that solutions being built for India’s resource-constrained health system could have wider relevance across the Global South, where many countries face similar shortages.
“The developing markets understand the challenge, they know that there’s so much burden on them and now they are more proactive in adopting AI,” he said.
Dr Devi Shetty, founder of Narayana Health and a prominent cardiac surgeon in India, told CNA that tech and AI’s potential lies in allowing scarce doctors and nurses to spend less time on routine work and more on patients.
He pointed out that nurses spend far more time with patients than doctors do, yet much of their day can be consumed by repetitive work such as taking routine measurements and recording information.
“The most boring, mundane job – now that job is best left with technology,” Shetty said.
For AI in healthcare to scale, there’s also the question of building trust in the technology, experts said.
Finding patients faster does not mean they will seek care sooner, said Rohan Devidas, who leads Qure.ai’s patient navigators in Goa. He recalled a 54-year-old man whose X-ray was flagged for a lung nodule, but the person saw little reason to undergo further tests as he felt healthy.
When calls fail to persuade patients, Devidas said his team turns to government health workers who know the community to explain the findings and encourage follow-up.
Some patients who initially resisted testing returned after such visits, he said.
Devidas said this was why frontline workers remain central even as AI becomes more capable.
“It is very important to have an interpersonal community connection with the patient,” he said. “Initially, it might be difficult for them to trust new technologies.”
Source: CNA/ss(cf)
Sign up for our newsletters
Get the CNA app
Stay updated with notifications for breaking news and our best stories
Get WhatsApp alerts
Join our channel for the top reads for the day on your preferred chat app
Get bite-sized news via a new
cards interface. Give it a try.
Click here to return to FAST
Tap here to return to FAST
FAST
