Researchers at the Indian Institute of Science (IISc), Bengaluru, are developing artificial intelligence tools to support early cancer screening and improve access to healthcare across India. Through the Translational AI for Networked Universal Healthcare (TANUH) initiative, the team is working on technologies designed to assist frontline health workers, particularly in areas with limited medical infrastructure.
The initiative focuses on making AI-based screening and clinical decision-support tools more accessible to India’s diverse population.
AI Tool Targets Early Oral Cancer Detection
One of TANUH’s key solutions, Aarogya Aarohan, helps frontline health workers screen for oral cancer and precancerous lesions. The system uses photographs captured through a mobile phone to assess suspicious areas inside the mouth.
The tool is particularly relevant in India, where tobacco, paan and gutkha use contribute to the burden of oral cancer. According to Professor Phaneendra K Yalavarthy, who leads TANUH’s work, the programme has screened more than 70,000 people.
The technology is designed to help identify individuals who may need further medical evaluation, rather than replace a clinical diagnosis.
AI Aims to Improve Breast Cancer Risk Assessment
The team is also developing an AI-powered tool to estimate breast cancer risk through a questionnaire that adapts its questions according to a person’s earlier answers.
The system asks up to 10 questions to classify individuals as being at higher or lower risk. Those identified as potentially high-risk can then be referred for mammography and further assessment.
Yalavarthy told The Indian Express that the tool showed about 83% accuracy compared with mammography in the team’s assessment. The result is a risk-screening measure, not a definitive cancer diagnosis, and further medical evaluation remains important.
Screening Beyond Cancer
TANUH’s research extends to other conditions, including diabetes, cardiovascular disease, kidney and liver diseases, and gestational diabetes.
Researchers are exploring whether images of the retina, captured using a specialised camera, can help flag people who may require additional tests for certain vascular and metabolic conditions. They are also developing AI methods to predict gestational diabetes earlier in pregnancy.
For mental health and neurological conditions, the team is investigating whether voice-based analysis can support the identification of potential warning signs.
These tools aim to help healthcare workers decide who may need additional testing or referral.
IndiNeuroFM: AI Designed for the Indian Brain
A major project is IndiNeuroFM, developed with IISc’s Centre for Brain Research. The initiative aims to build an India-specific, multimodal AI model for neurological healthcare.
Many existing medical AI systems are trained predominantly on data from other populations. Differences in disease patterns, nutritional conditions, imaging equipment and clinical resources can affect how well those systems perform in Indian settings.
IndiNeuroFM is intended to account for these differences. For example, it could help flag findings such as a possible skull fracture or brain bleed on a head CT scan, supporting doctors who work in areas where specialist radiologists are scarce. The model is intended to assist clinical assessment, not make treatment decisions independently.
Building a Shared AI Healthcare Platform
TANUH is also developing platforms to help researchers and healthcare providers test, deploy and monitor AI solutions. Its BODH platform supports the evaluation of healthcare AI models using Indian datasets, while SAMVIT is designed to help integrate screening tools into existing health systems.
The broader goal is to move promising technologies beyond isolated research projects and towards practical use across hospitals and community healthcare programmes.
Making Healthcare AI More Accessible
AI tools could help frontline workers identify people who need further investigation, especially in regions with shortages of specialists and diagnostic facilities. However, their effectiveness depends on reliable data, independent validation, privacy safeguards and appropriate clinical oversight.
By developing tools around India’s population and healthcare conditions, IISc and TANUH are working towards a model in which AI supports earlier detection and more accessible care while doctors remain central to diagnosis and treatment.