Duke Uses AI, Robots to Pinpoint 15-Microbe, 6-Fiber Mix for Stable Butyrate

3 articles · Updated · dongascience.com · Jul 28

Five experimental rounds let Duke researchers test up to 390 conditions at a time, using AI and robotic automation to identify probiotic-fiber combinations that maximize butyrate, a gut-health metabolite.
Trillions of possible combinations made conventional modeling impractical, so the team paired machine learning and optimization algorithms with high-throughput lab experiments to narrow the search.
Inulin, Bacteroides uniformis, Anaerostipes caccae and Prevotella copri emerged as a robust butyrate-producing combination that stayed effective even after extra microbes or compounds were added.
The Nature Chemical Biology study shifts the goal from simply adding 'good' bacteria to designing the gut environment around them, with mouse tests for inflammatory bowel disease now underway.
Researchers said the approach could support personalized probiotic-fiber formulations, but it still must prove it works reliably in the far more complex human gut.