An Intelligent Intestine-on-a-Chip for Rapid Screening of Probiotics with Relief-Enteritis Function.

Abstract

Screening probiotics with specific functions is essential for advancing probiotic research. Current screening methods primarily use animal studies or clinical trials, which are inefficient and costly in terms of time, money, and labor. An intelligent intestine-on-a-chip integrating machine learning (ML) is developed to screen relief-enteritis functional probiotics. A high-throughput microfluidic chip combined with environment control systems provides a standardized and scalable intestinal microenvironment for multiple probiotic cocultures. An unsupervised ML-based score analyzer is constructed to accurately, comprehensively, and efficiently evaluate interactions between 12 Bifidobacterium strains and host cells of the colitis model in the intestine-on-a-chips. The most effective contender, Bifidobacterium longum 3-14, is discovered to relieve intestinal inflammation and enhance epithelial barrier function in vitro and in vivo. A distinct advantage of this strategy is that it can intelligently differentiate small therapeutic variations in probiotic strains and prioritize their efficacies, allowing for economical, efficient, accurate functional probiotics screening.