IIT-M, CMC develop AI tools to catch kidney disease early
A 2025 ICMR-INDIAB study of 25,408 people found impaired kidney function — meaning the kidneys were not filtering blood as effectively as normal — in 3.2% of participants. A 2025 study of 3,350 agricultural workers in Tamil Nadu found CKD in 5.31% of participants. The model was developed using a publicly available dataset of 400 patient records, including 250 classified as having CKD and 150 without the disease. It assessed 26 clinical and laboratory variables. In cases studied, kidney volumes ranged from about 120 mL to 245 mL, while tumour volumes ranged from 2 ml to 24 ml. Tumour burden ranged from about 1% to 10.6%.
Researchers tested four computer methods and selected a “random forest” model, which combines the results of many simple computer-generated decisions to identify patterns linked to disease. “We aim to develop a digital twin of the kidney that can be trained on patient data to model disease progression and help predict how the condition may evolve,” said professorG L Samuel of the Department of Mechanical Engineering at IIT Madras. “The next step is to obtain more patient data and validate the model with clinical cases,” Samuel said. “Validation may take about two years, while ethical clearance and hospital implementation could take about five years.”
Chennai: Artificial intelligence tools that could help doctors assess kidney disease risk, interpret CT scans, and measure kidney tumours have been developed by researchers at IIT Madras in collaboration with doctors at Christian Medical College, Vellore. Chronic kidney disease, or CKD, is the loss of the organ’s ability to filter waste and excess fluid from the blood. Since CKD can go unnoticed in its early stages, some patients are diagnosed only after considerable damage has occurred. In advanced cases, they may need dialysis, in which a machine filters the blood, or a kidney transplant. Scientists at IIT Madras and CMC Vellore have produced three tools, each aimed at a different stage of diagnosing and assessing kidney disease. The first uses clinical and laboratory information to estimate a person’s risk of CKD. The team also built a prototype interface intended to make the system easier for doctors to use. The third tool reconstructs a three-dimensional image of a kidney and tumour from CT scans. It calculates tumour volume and burden. “If a patient’s scan and clinical data are fed into the digital twin, it could help doctors understand how kidney disease may progress over three months, six months or a year, particularly if it is not treated. The tools remain at the research stage and cannot yet predict an individual patient’s outcome or replace a doctor’s judgement.