IIT Madras and CMC Vellore build AI tools for early kidney disease detection

This collaboration is a good example of where current AI is already delivering meaningful progress without needing frontier-scale models. IIT Madras and CMC Vellore are combining conventional machine learning and deep learning with domain expertise to address a high burden disease in India, where chronic kidney disease is often diagnosed late. The tools span risk prediction, image classification and 3D reconstruction, hinting at a future in which a patient’s lab values and scans feed into a continuous, AI-augmented picture of organ health.

From an AGI perspective, these systems are narrow and supervised. They do not “understand” medicine in any general way, but they showcase how compositional use of models across modalities can deliver something like a proto digital twin for an organ. That pattern, repeated across specialties, nudges healthcare towards environments where more of the relevant state is digitized and modeled. Those environments are fertile ground for more agentic systems that can coordinate tests, suggest interventions and simulate outcomes.

The global significance is that this work is happening in India’s public research ecosystem, not just in US or Chinese Big Tech labs. As more high quality, locally relevant datasets and models emerge from places like IITs and teaching hospitals, we can expect a more multipolar landscape in applied medical AI, even if the very largest general models remain concentrated among a few global firms.