IIT Madras and CMC Vellore pioneer AI tools for kidney disease detection

New technologies aim to predict risk, classify abnormalities, and measure tumour burden.

Kidney disease often develops quietly, with symptoms appearing only after significant damage has occurred. In response, researchers from the Indian Institute of Technology Madras (IIT Madras) and Christian Medical College (CMC), Vellore, have unveiled three artificial intelligence (AI) tools designed to support earlier detection and more precise assessment of kidney conditions.

Led by Prof. G.L. Samuel of IIT Madras and research scholar Jennifer Delighta, in collaboration with Prof. Santosh Varughese of CMC Vellore, the team has focused on three critical aspects of kidney care. The first innovation is a machine-learning model that predicts chronic kidney disease (CKD) risk using clinical and laboratory data. The team has already developed a prototype interface to make the tool accessible and interpretable for clinicians.

The second breakthrough is a deep-learning classifier trained on more than 12,000 CT images. This system can distinguish between four categories: normal kidneys, cysts, stones, and tumours. The team presented its work in CT-based multiclass classification at the World Congress of Nephrology, highlighting its potential to streamline radiological analysis.

The third technology takes imaging a step further by creating patient-specific 3D kidney models from CT scans. These models allow doctors to estimate tumour volume and assess the percentage of kidney tissue affected, offering a more comprehensive view of disease progression.

According to Prof. Samuel, the goal was to build intelligent systems that combine machine learning with clinical expertise to help doctors make faster, better-informed decisions. Jennifer Delighta emphasized that early detection is crucial, and these tools could significantly improve treatment planning by identifying at-risk patients sooner.

The researchers envision these innovations contributing to a “kidney digital twin”- a virtual patient-specific model that could track disease progression, forecast changes, and guide personalized treatment strategies. While the promise is clear, the technologies are still in development and require validation with larger patient datasets before they can be integrated into routine hospital practice.

Globally, the World Health Organization estimates that 674 million people live with CKD, underscoring the importance of early detection. Current KDIGO guidelines recommend prioritizing screening for individuals with diabetes, hypertension, and cardiovascular disease. The IIT Madras-CMC Vellore collaboration represents a significant step toward using AI not as a replacement for clinical judgment, but as a powerful supplement to enhance kidney care.