ASHA workers, the backbone of India’s primary healthcare system, will soon add another crucial task to their house calls. As the first point of contact for healthcare in remote villages and towns, these workers already handle tasks like vaccination and health reviews. Now, they will also conduct detailed conversations with families about their personal health status and family history of diseases, guided by a formatted questionnaire.
The goal is to collect vital data that can help AI tools identify breast cancer risk factors and patterns specific to Indian women. This initiative is part of an ambitious project by AIIMS Delhi aimed at improving early breast cancer detection and reducing mortality, which remains among the highest globally.
The project is one of the first under the government’s flagship scheme to establish three Centres of Excellence (CoE) in AI, focusing on healthcare, agriculture, and sustainable cities. AIIMS has secured ₹300 crore in funding to lead the healthcare CoE in partnership with IIT Delhi, working on various national programs to address non-communicable diseases.
“Unfortunately, the majority of breast cancer cases that we’re picking up in India are still stage three or stage four, which are difficult to treat. AI will help us on two fronts. First, it can interpret the data pool and identify which women may need mammograms — specialised X-rays to check for signs of breast cancer — while reassuring others with low-risk profiles. This can facilitate early screening,” says Dr Krithika Rangarajan, associate professor of radiology, Institute-Rotary Cancer Hospital (IRCH), AIIMS, Delhi.
Second, the AI tool can read mammograms correctly, picking up even the smallest signs of cancer. “This allows advanced breast cancer screening in remote areas where there is a shortage of trained radiologists. Otherwise, a lot of time is lost in diagnosis,” she adds.
How will the AI model work?
The AI model will first scan a five-year database of women tested for breast cancer at AIIMS, NCI Jhajjar and PGI Chandigarh regardless of whether they were found clean or had developed cancer. Analysing them, it will predict the risk of developing breast cancer by combining a patient’s general test results with their lifestyle and family history data. This would then allow doctors to sift those at higher risk and recommend interventions.
The first part of the study involves identifying Indian risk factors for breast cancer. “For this, we will be using ASHA workers to collect the data, which will be fed into the AI tool. It will extract common risk factors and then recommend mammograms for women it thinks are cancer-prone. It will also help codify what constitutes a no-risk category,” says Dr Rangarajan.
Story continues below this ad
The second part will include AI’s analysis of mammography based on the risk factors found in participants. “The AI-trained systems can learn to recognise complex features in mammograms that indicate cancer. So no radiologist will be required on-site. AI can, therefore, help reduce the amount of manual work involved in screening, which can lower costs,” says Dr Rangarajan.
Mammography is a very complex modality and a radiologist requires many years of experience to be able to read it correctly. “Can we now, with the AI assistance, allow a relatively less trained person to read the mammography? That’s what we are working on as it would enable a larger number of women to get mammograms,” she says. According to Globocan, breast cancer comprises 10.6 per cent of all cancer deaths in the country and this facility could certainly reduce those numbers.
When will a large-scale rollout be possible?
The pilot study has a certain set of questions, which will get updated and refined over time. “Eventually the goal is to bring such a screening tool to the national health programme and integrate it in the workflow of all ASHA workers,” says Dr Rangarajan. In studies and trials globally, AI has been able to analyse genetic and molecular data to create even personalised treatment strategies for individual patients.
“There will be several components of developing these tools, clinical studies and implementation research over the next four years,” says Dr Rangarajan. If the AI tool is found to be effective, then an economic assessment will be conducted on the cost required to scale this up pan-India and take it up for licensing. They can then be brought to the market through a company or NGO.
Story continues below this ad
Data protection
What about data protection, considering they cover a huge swathe of the population? “Data security will be our top priority and we will be building a mechanism so that our data is secured and encrypted. We will also begin the study without disrupting the way the patient is treated. So you are using AI for your analysis. There will be guardrails as well as both human and expert oversight,” assures Dr Rangarajan.
