IIT Madras and CMC build AI to catch kidney disease early
AI could help spot kidney disease earlier. IIT Madras and CMC Vellore have unveiled three tools for risk prediction, CT scan triage and 3D tumour assessment.
AI kidney disease detection is moving closer to the clinic. Researchers at IIT Madras and Christian Medical College (CMC), Vellore, have developed three AI-based tools designed to help identify and assess kidney conditions earlier in the care journey.
Announced on September 3, 2026, the collaboration brings together machine learning, medical imaging and 3D reconstruction to give clinicians faster and more standardised information. The broader goal is to identify kidney problems earlier, potentially helping patients avoid disease progression and costly treatments such as dialysis.
The three tools address different stages of the process. The first uses routine clinical and laboratory data to predict a person's risk of chronic kidney disease. The second uses deep learning to classify CT scans as showing a normal kidney, cyst, stone or tumour.
The third is an open-source 3D imaging platform that reconstructs kidneys from CT scans and measures tumour and kidney volumes.
3 Tools for three parts of the diagnosis
The kidney disease risk model was tested across multiple machine learning algorithms before the researchers selected a random forest approach. It was initially developed using a public dataset of about 400 records containing 26 clinical and laboratory variables associated with chronic kidney disease.
The imaging system takes a different approach. It was trained on approximately 12,400 publicly available CT images and is designed to automate initial classification across four categories: normal kidney, cyst, stone and tumour.
The third tool adds a 3D view. It reconstructs kidneys from CT scans and calculates kidney and tumour volumes, providing information that could support surgical planning and help doctors monitor disease over time.
The project is led by Prof G L Samuel and researcher Jennifer Delighta at IIT Madras, in collaboration with Prof Santosh Varughese, a nephrologist at CMC Vellore.
The researchers see the three tools as a foundation for a future kidney “digital twin”, essentially a virtual representation of a patient's kidney that could combine clinical and imaging information to model disease progression and support more personalised decisions.
From research to real-world care
The technology is still at the development stage, and wider clinical use is not expected immediately. Currently, the institutions plan to test the systems using multi-centre clinical datasets and explore integration with wearable sensing platforms for longer-term monitoring.
Broader validation and hospital deployment could take several years as researchers collect more patient data and work through ethical and regulatory requirements. The intended use is particularly relevant to community and primary-care settings, where early identification could help determine which patients need a specialist nephrology referral or further investigation.
If validated, the tools could also help standardise CT interpretation in resource-constrained settings and provide a more consistent way to track tumour response during treatment. But building reliable medical AI requires more than developing the algorithms.
Access to large, high-quality patient datasets remains a challenge in India, limiting how quickly these systems can be trained and tested across different populations. The IIT Madras-CMC team is therefore looking for additional partners and datasets as it works towards broader validation.
Right now, this project represents an early step towards using AI not just to detect kidney disease, but to build a more complete digital picture of how it develops and responds to treatment.