Predicting Radiation Exposure in Mice Using a Classification Model Based on Gut Microbial Alterations

Abstract

Predictive modeling is a technique used to analyze data and forecast future outcomes by identifying patterns and relationships.  In microbiome research, predictive modelling has primarily been applied to predict diseases, which can be leveraged for precision medicine. However, microbiome-based models focused on predicting environmental exposure appear less common. In particular, we were concerned with long-term space exposure and its effects on host gut microbiome interactions, provided it has wide implications for human health. As a consequence of space exposure, disruptions to the microbiome can compromise immune function and increase susceptibility to pathogens, posing significant health risks to astronauts. Given that the space environment contains multiple stressors that may alter the gut microbiome, we developed a predictive model specifically for exposure to space-type radiation based on gut microbe abundance. The objective was to identify predictor taxa from the gut microbiome of laboratory-irradiated mice and utilize their presence and abundance to predict radiation exposure in a space-exposed mouse model. Our predictive model encompassed nine taxa indicative of space-type radiation which were identified using core microbiome and indicator species analyses. However, none of these nine predictors appeared in our space-exposed testing dataset, nor were we able to validate the taxonomic model. Thus, our model needs to be validated and tested further. These findings underscore the challenges in using microbial data to model radiation.