AI model predicts kidney stone type from urine tests
Mayo Clinic researchers created an artificial intelligence system to forecast kidney stone makeup using standard urine test results in a new study.
The new tool correctly identified stone composition in 73% of patients. Older calculation methods alone only reached 33% accuracy during patient care visits. This improvement helps medical staff understand stone types without needing the physical stone for analysis.
Developers built the system using information from over thirteen thousand individuals. They used age details and sex data alongside specific urine test outcomes. The software learned to recognize complex patterns within the collected medical records before testing its accuracy on new data.
They later checked how well the model worked by testing it against results. They evaluated the system using an additional group of one thousand six hundred fifty three people. The evaluation showed the artificial intelligence system worked better than the traditional math method for all age and sex groups combined.
Understanding the exact material of a stone allows doctors to choose better treatments. This process creates plans that lower the chance of future stones forming again.
Many people never pass stones that can be examined for testing. This technology offers a way to get useful answers from existing lab records.
“AI can identify those patterns and could help us get more clinical value from data we are already collecting.”

