A Medical Digital Twin for the Gut Microbiome - Leo Grady Seminar

Leo Grady, an internationally recognized industry leader in AI for health, explored the applications of large language models in understanding the gut microbiome's impact during a seminar held by the Oden Institute of Computational Engineering and Sciences. The special seminar, hosted by Charley Taylor, Director of the Center for Computational Medicine, was held on March 6, 2025. Grady’s presentation surrounding the work of his newest company, Jona, engaged an audience of faculty and students from the Oden Institute, alongside clinicians from Dell Medical School. 

The gut microbiome has been demonstrated to have broad implications in human health and disease, showing correlations with several mental health disorders, autoimmune diseases, and digestive disorders. The efficacy of certain medications, fertility, and human longevity may also be underpinned by the complex web of microorganisms that call our digestive tracts home. While research surrounding the microbiome’s impact on specific diseases is moving at a breakneck pace, the new prevalence of data is often difficult for clinicians and patients to put into practice, in the form of individual probiotic supplements, medications, or dietary and lifestyle changes. 

“This is where Jona comes in,” said Grady. The company created a Large Language Model (LLM) which is trained exclusively on medical and scientific literature sourced from the PubMed Database.

As the AI tool extracts analytic and interventional knowledge from evidence based sources, it’s bridged with Next Generation DNA Sequencing techniques, which reveal the makeup of an individual patient’s gut flora.

“This allows us to sort through a huge volume of data. Our LLM is able to rank studies in terms of strength and efficacy, allowing us to provide our users with a medical digital twin that represents their gut microbiota - a summary of links to certain diseases based on elevations or depressions of certain flora, and a unified set of interventional recommendations shown to reduce these links in terms of supplements, diets, and lifestyle modifications,” Grady said. 

Grady’s interest in exploring the implications AI for the gut microbiome might have stemmed from watching several family members interface with diseases that were difficult to identify and manage clinically.

“Many members of my family have struggled immensely with Colitis, Crohn's disease, Celiac disease, and other conditions with strong links to the gut microbiome,” he shared. His family members often took years to receive a definitive diagnosis and found themselves inundated by advertisements from companies aiming to sell probiotic supplements, without a strong basis in evidence-based research, despite critical advancements in the field. These challenges inspired Grady to leverage his background in AI and healthcare, in order create a product with potential to bridge a critical need within and beyond gastrointestinal care.

The application of digital twin technology sparked some discussion among seminar attendees, who wondered about geographical and temporal differences between gut flora found in different patient populations.

As much of the data on gut health is sourced from studies in a wide range of countries, there were questions about how to establish a reference point for what qualifies as an elevation or depression of a specific bacterial strain. Many attendees were also interested in understanding how the LLM controls for certain variables, including the cohort size of certain interventional studies. 

As artificial intelligence and healthcare become increasingly intertwined, Grady's seminar brought new perspective into an ongoing conversation at the Oden Institute.