“For us, this approach has the potential to transform the prevention and management of urolithiasis, making health care more precise and personalized,” says Juan Fulla, MD, MSc.
In this interview, Juan Fulla, MD, MSc, shares the background and key findings from the study, “Artificial Intelligence in Urology: Application of a Machine Learning Model to Predict the Risk of Urolithiasis in a General Population,” for which he served as the senior author. Fulla is the chief of the robotic program and an associate professor of urology at the University of Chile in Santiago, Chile.
Video Transcript:
Could you describe the background/rationale for this study?
Our study was motivated by the growing need to enhance predictive tools in medicine, specifically in the area of urolithiasis, where kidney stones pose a significant public health burden in the United States, and also here in Chile. Until now, the use of artificial intelligence in this area has primarily focused on the identification and characterization of the stones, but through diagnostic imaging. However, there was, and we found, that there is a notable gap in AI applications that consider a combination of patient-specific characteristics, [such as] lifestyle and also diet. For that reason, we aim to fill this gap with a comprehensive approach. For that reason, we gather a robust sample of nearly 1000 Chilean participants and analyzed the demographic, lifestyle, and health data through an exhaustive questionnaire. Finally, our goal was to develop a predictive model that not only identifies individuals at risk of kidney stone formation with high accuracy, but also provide a personalized recommendation based on modifiable factors. For us, this approach has the potential to transform the prevention and management of urolithiasis, making health care more precise and personalized.
What were the key findings?
The findings of our study were quite revealing. We achieved an 88% accuracy in identifying individuals at risk of kidney stone formation using various classifiers, including logistic regression, decision trees, random forest [plots], and extra trees. We found several protective factors. They were highlighted by the algorithm and were crucial, such as adequate hydration, regular physical activity, and healthy dietary patterns, and specifically higher intake of fruits and vegetables. Balanced dietary consumption and careful selection of the protein sources correlated with a lower risk of kidney stone formation. On the other hand, we also identified several risk factors, including gender disparities, where men were found to be 2.3 times more likely to develop kidney stones than women. The perception of being thirsty during the day or observing dark during the during the day were strong predictors of risk. We think these findings underscore the importance of considering a variety of individual factors in the prediction of and prevention of urolithiasis.
This transcription has been edited for clarity.