The Indian Institute of Science’s Translational AI for Networked Universal Healthcare (TANUH) centre is deploying artificial intelligence tools for early detection of non-communicable diseases, with a focus on oral and breast cancer screening by frontline health workers. Established as one of four government-designated AI Centres of Excellence, TANUH operates as a non-profit developing diagnostics and decision-support tools for primary care settings across India.
For oral cancer, TANUH’s Aarogya Aarohan mobile application enables health workers to photograph lesions and identify pre-cancerous conditions using AI. The tool has screened more than 70,000 patients, with deployment concentrated in high-incidence regions including Varanasi and Mathura in Uttar Pradesh, Dakshina Kannada in Karnataka, and Thanjavur in Tamil Nadu. India’s oral cancer burden is driven largely by tobacco, paan, and gutkha use.
Breast cancer screening addresses a different gap: only about 3,500 mammogram machines serve roughly 200 million Indian women aged 40 and above, covering approximately 1% of the target population. Indian women also develop breast cancer earlier, with peak incidence between ages 40 and 50 compared to 50-60 in Western countries. TANUH’s reinforcement-learning-driven questionnaire asks up to 10 adaptive questions to stratify risk, achieving 83% accuracy against mammography with a 97% negative predictive value.
The centre’s retinal vascular screening tool uses a handheld fundus camera to capture eye images without pupil dilation, flagging diabetes, cardiovascular, liver, and kidney disease risk while estimating biochemical parameters such as eGFR for kidney function. An anti-fraud tool for the PM-JAY insurance scheme analyses kidney ultrasound images to verify dialysis necessity, addressing an estimated 30% of the scheme’s budget spent on dialysis.
In neurology, TANUH collaborates with IISc’s Centre for Brain Research on IndiNeuroFM, a sovereign multimodal foundation model for the Indian brain trained on data from over 20,000 patients with a target of 60,000. The model addresses India-specific conditions like neurotuberculosis and nutritional deficiencies, and accommodates the prevalence of 1.5 Tesla MRI scanners versus 3 Tesla systems common in Western training data.
Two platform initiatives support broader deployment: BODH provides a benchmarking platform for testing AI models on Indian datasets, while SAMVIT hosts TANUH’s screening tools and enables integration of third-party solutions for community health worker programmes. Both are designed as digital public infrastructure, with tools built in-house using data from approximately 50 clinical partners and made available as digital public goods.
TANUH partners with the Ministry of Education, National Health Authority, PM-JAY, Ministry of Health, ICMR institutes, state health departments, and over 60 hospitals and research institutions. Professor Phaneendra K Yalavarthy, TANUH’s Chief Project Manager and a professor in IISc’s Department of Computational and Data Sciences, identifies the SAMVIT platform as the centre’s biggest bet for sustainable, scalable AI deployment in Indian healthcare.
