Year in Review: Artificial Intelligence and Nephrology

The future of kidney care looks bright — according to a new review article published by UF Intelligent Clinical Care Center (IC3) researchers in the journal Nature Reviews Nephrology.

IC3 and PRISMAp Director, Azra Bihorac, M.D. MS FCCM FASN, and IC3 Associate Director, Benjamin Shickel, Ph.D, took a look back at 2023 to identify how artificial intelligence developments revolutionized the field.  According to their article, nephrology experienced a significant evolution — spurred on by text-based algorithms called large language models and patient data from multiple sources.

bihorac shickel
Azra Bihorac and Benjamin Shickel

Most Americans became familiar with AI advancements through the release of the online (LLM) ChatGPT at the end of 2022 — which made machine learning accessible to non-technical users worldwide for the first time. However, Bihorac and Shickel believe that text is just one part of the modern health ecosystem.

“Creating ChatGPT was an important first step and LLM technologies like it have the potential to shape future research and augment clinical practice in nephrology,” Shickel said, “But the past year saw an important turning point in the evolution of artificial intelligence utilizing other forms of data.”

According to the article, the field of nephrology has experienced notable advancements in AI within just the past year. Recent advancements have enabled AI to analyze non-traditional sources of data, such as photographs of the eye and blood flow sounds, to detect biomarkers for kidney disease and stenosis. Additionally, a new automated quantitative approach for measuring the structure and characteristics of tissues, termed pathomics, was found to be a reliable predictor of kidney disease progression.

“These innovations offer opportunities for precision medicine and non-invasive screening for kidney disease,” Bihorac said. “By using multimodal AI techniques that leverage diverse patient data sources, such as images, audio, video and biological signals, we can unlock the next generation of precision nephrology.”

In 2024, Bihorac and Shickel expect that artificial intelligence will continue to transform nephrology as a whole.

According to the pair, personalized patient data from various sources will be combined with advanced LLMs to help develop foundational medical models. These models will revolutionize the field of nephrology by providing a natural language interface, advancing both medical knowledge and reasoning abilities.

“This approach has the potential to propel nephrological medicine into the next digital age,” Bihorac said. “We are excited to see how the field continues to evolve and encourage innovation.”