Last Updated on July 27, 2026 by Staff
The human gut is home to one of the complex ecosystems on Earth. Trillions of microorganisms—including bacteria, fungi and other microbes—live inside the system and play essential roles in maintaining overall health. These microbes help digest food, produce nutrients, strengthen the immune system and even influence brain function.
However every person’s gut microbiome is unique. Factors such as diet, lifestyle, medications, age and environmental exposure constantly shape which microbes are present and how they interact. This diversity makes it extremely difficult to develop probiotics and prebiotics that work consistently for everyone.
Researchers at Duke University have now developed a method that combines artificial intelligence, robotics and laboratory experiments to create more effective probiotic and prebiotic combinations. Their study, published in Nature Chemical Biology could help transform the growing gut health industry by making treatments more predictable and personalized.
Smarter Design
Traditional probiotics introduce bacteria into the digestive system while prebiotics provide nutrients that help those bacteria survive and grow.
The challenge is that simply adding bacteria does not guarantee success.
As planting a seed in poor-quality soil may prevent it from growing, beneficial microbes often struggle to establish themselves in the crowded and highly competitive environment of the human gut.
According to lead researcher Professor Ophelia Venturelli Improving gut health requires engineering both the microbes and the environment they live in.
The team’s goal was to identify combinations of bacteria and dietary fibers that work together to produce health benefits regardless of differences in individual gut microbiomes.
Of testing random combinations they developed a systematic approach capable of exploring an enormous number of possibilities with remarkable efficiency.
AI and Robots
The research combined modeling with automated laboratory robotics and machine learning.
Scientists selected 15 gut bacterial species and six dietary fibers known to influence the production of butyrate—a short-chain fatty acid essential for maintaining healthy intestinal cells.
With only 21 variables the number of possible combinations reached trillions making manual testing impossible. To solve this problem researchers used a machine-learning technique called Bayesian optimization. This method intelligently selected the informative experiments helping the computer model learn faster while simultaneously searching for the best bacterial and fiber combinations.
Laboratory robots then carried out hundreds of experiments in parallel testing up to 390 conditions at the same time. Over five rounds of automated testing the system continuously improved its predictions by combining experimental data with computer-generated models.
The researchers discovered complex microbial interactions that would have been almost impossible to predict using conventional laboratory methods.
Winning Formula
One successful combination stood out during the experiments. Researchers found that the dietary fiber inulin worked well alongside two bacteria—Bacteroides uniformis and Anaerostipes caccae—plus a third species called Prevotella copri.
Together these microbes consistently produced levels of butyrate, one of the most beneficial compounds generated by healthy gut bacteria.
Butyrate serves as an energy source for intestinal cells, reduces inflammation and helps maintain the protective lining of the digestive tract.
More encouragingly, this microbial partnership continued producing butyrate effectively even when additional bacterial species and changing environmental conditions were introduced. This suggests the combination may remain reliable despite the variation found between different peoples gut microbiomes.
Such stability has long been one of the challenges facing probiotic research.
Future Medicine
The researchers are already testing this promising combination in mouse models of inflammatory bowel disease with early results appearing encouraging. Beyond this discovery the team believes their AI-driven workflow can rapidly identify many other customized probiotic-prebiotic combinations for different gastrointestinal disorders.
The global probiotic and prebiotic market is currently worth approximately $130 billion. Many commercial products still struggle to demonstrate consistent scientific effectiveness.
By using intelligence, robotics and high-throughput laboratory experiments together future products could be tailored to individual health conditions rather than relying on generalized formulations. This approach may eventually lead to treatments for inflammatory bowel disease, irritable bowel syndrome, digestive disorders and other microbiome-related illnesses.
Broadly the study demonstrates how combining AI with automated experimentation can dramatically accelerate biomedical discovery.
After spending years testing countless possibilities manually, researchers can now intelligently search enormous biological design spaces uncovering microbial partnerships that were previously hidden and bringing the next generation of precision gut health therapies much closer to reality.