Active Machine learning for formulation of precision probiotics.

Area of Science:

• Microbiome science
• Pharmacology
• Machine learning applications

Background:

• The human gut microbiome is crucial for health, and dysbiosis is linked to disease.
• Precision probiotics are being developed to restore microbial balance.
• Effective delivery of probiotics to the distal gut is essential for their efficacy.

Purpose of the Study:

• To predict the impact of pharmaceutical excipients on the intestinal proliferation of Lactobacillus paracasei.
• To explore the application of active machine learning (ML) in optimizing probiotic formulations.
• To identify excipients that enhance probiotic survival and proliferation in the gut.

Main Methods:

• Utilized active machine learning (ML) with uncertainty sampling.
• Started with a small dataset of 6 bacteria-excipient interactions.
• Predicted the effects of 111 additional excipients on Lactobacillus paracasei proliferation.

Main Results:

• The active ML model achieved an average certainty of 67.70%.
• Experimental validation confirmed 75% accuracy in predicting excipient-probiotic interactions.
• Successfully predicted the effects of 111 new excipients.

Conclusions:

• Active ML can effectively predict excipient effects on probiotic proliferation.
• This approach enables the development of superior probiotic delivery systems.
• This marks the first application of active ML in microbiome science for formulation development.