University of Missouri researchers are using artificial intelligence to uncover hidden clues among the trillions of bacteria living in the human gut, including changes that could one day help scientists detect diseases earlier.
The Mizzou team — which brings together expertise in medicine, data science and engineering — developed a new machine-learning tool capable of finding biological signals that conventional methods miss.
The tool, called Metric Learning for Statistical Inference, or MeLSI, cuts through the complexity of microbiome data to identify the microbes driving key biological changes. That insight matters because researchers increasingly believe microbiome changes can begin months or even years before symptoms appear.
“We’re starting to see that small changes in the gut microbiome may occur long before people begin to show symptoms of neurological diseases such as Alzheimer’s,” said Ai-Ling Lin, a professor of biological sciences in Mizzou’s College of Arts and Science , professor of radiology in the School of Medicine and a NextGen Precision Health investigator. “If we can identify those changes earlier, we could find ways to mitigate or even reverse them.”
Traditional analytical methods can reveal broad differences between healthy and diseased microbial communities. MeLSI goes a step further by identifying the specific microbes responsible for those changes, giving researchers a clearer picture of what’s changing within the microbiome.
“Standard approaches are like using a generic ruler to measure everything,” said Nathan Bresette, a doctoral student in Mizzou’s School of Medicine and lead author of the study. “Instead, MeLSI works more like a smart scanner that learns to ignore background clutter and focus on what matters most.”
When Mizzou researchers tested MeLSI using data from several previously published studies, the tool consistently revealed meaningful patterns that conventional methods missed.
In a study of more than 1,100 adults, MeLSI didn't just confirm that men's and women's gut microbiomes differ. It pointed to the specific bacteria most responsible for that difference, giving researchers a concrete starting point for follow-up work. In a separate study comparing people on Western and high-fiber diets, MeLSI picked up a real dietary effect on the microbiome that standard methods missed.
Together, the findings demonstrate how the Mizzou-developed tool could give researchers a powerful new way to study how the microbiome influences health and disease, potentially leading to earlier detection and more targeted interventions.
The study, “MeLSI: Metric Learning for Statistical Inference in microbiome community composition analysis,” was published in mSystems. Co-authors are Aaron Ericsson and Carter Woods at Mizzou.
mSystems