IISc centre develops AI tools for early cancer and disease screening

Translational AI for Networked Universal Healthcare, or TANUH, operates as an AI Centre of Excellence at the Indian Institute of Science in Bengaluru. Prof Phaneendra K Yalavarthy serves as the Chief Project Manager of the centre and a professor at the Department of Computational and Data Sciences. The Indian Government established TANUH along with three other centres focused on education, agriculture, and sustainable cities to develop sector-specific solutions at scale. Non-communicable diseases account for roughly seventy percent of the current disease burden in India. The centre builds applications for frontline health workers like nurses and ASHA workers rather than patients or doctors directly.

Screening Tools

TANUH focuses on oral and breast cancer screening, alongside retinal and vascular diseases, gestational diabetes, mental health, and neurology. The World Health Organization identifies oral and breast cancer as requiring population-level screening. For oral cancer, the centre uses a mobile phone app called Aarogya Aarohan to take pictures of lesions caused by tobacco, paan, or gutkha use. This tool has screened more than 70,000 patients and targets regions with high incidence rates such as Varanasi and Mathura in Uttar Pradesh, Dakshina Kannada in Karnataka, and Thanjavur in Tamil Nadu. For breast cancer, India has about 20 crore women above the age of 40 but only around 3,500 mammogram machines, leaving only about one percent of the target population screened. Furthermore, the highest incidence of breast cancer in India occurs between the ages of 40 and 50, which is earlier than the 50 to 60 age range seen in Western societies. High-risk patients make up about four to five percent of the general population.

Risk Assessment

The centre built a reinforcement-learning-driven questionnaire with only 10 questions that changes dynamically based on previous answers to perform India-specific risk stratification for breast cancer. This tool achieves about 83 percent accuracy compared to a mammogram. The negative predictive value reaches 97 percent, while the positive predictive value sits at 60 to 65 percent. TANUH also works on retinal vascular screening and is developing a foundation model for the Indian brain in partnership with the Centre for Brain Research at IISc. Their product user interfaces are multilingual, intuitive, gesture-based, and voice-based to suit point-of-care field use by frontline workers.

Context

Non-communicable diseases present a major public health challenge across India due to changing lifestyles and an aging population. Digital health tools and artificial intelligence help bridge severe gaps in specialist availability by empowering primary healthcare workers in remote areas. Similar technology-driven screening initiatives are increasingly adopted worldwide to manage large patient volumes efficiently.