The gut microbiome plays a significant role in the health and development of infants, with dysbiosis or imbalances in the microbial community being associated with gastrointestinal diseases and neurodevelopmental deficits. Traditional laboratory experiments to understand these interactions are time consuming and challenging. Researchers at the University of Chicago have developed a new generative artificial intelligence (AI) tool called Q-net, which creates a virtual model of the infant microbiome. This digital twin predicts the changing dynamics of microbial species in the gut as the infant develops. Using data from fecal samples collected from preterm infants in the NICU, the model accurately predicted which babies were at risk for cognitive deficits with 76% accuracy.
The new approach using generative AI to build a digital twin of the microbiome allows researchers to model the interactions of bacteria as they change in a preterm infant’s gut. This bridging of computer science, engineering, mathematics, and life sciences replicates the behavior of biological systems in a way that traditional wet lab experiments cannot achieve. Q-net drastically speeds up the process of testing out interactions between bacterial species, making it possible to identify links to specific outcomes quickly. The model was trained using data from UChicago’s Comer Children’s Hospital and validated using data from Beth Israel Deaconess Medical Center in Boston, accurately predicting which babies were at risk for cognitive deficits.
The Q-net model also suggests that interventions like restoring the abundance of a particular bacterial species could reduce the developmental risk in about 45% of babies. However, the model also shows that incorrect interventions can worsen the risk, highlighting the importance of precise timing and the specific bacteria being introduced. Q-net can identify potentially interesting combinations of bacteria, narrowing the search for treatment targets in the proverbial haystack of the gut microbiome. Research partners like Erika Claud are testing potential interventions in bioreactors that simulate the live gut microbiome environment to observe the effects.
At its core, Q-net models large numbers of variables that interact with each other, making it a versatile tool that can be applied to other systems beyond the microbiome. It could potentially be used to study the evolution of viruses or social phenomena like public opinions. With a large amount of data, this system can be trained to figure out connections and capture subtle differences, making it applicable to a wide range of applications. The ability of Q-net to predict outcomes based on interactions between microbial species in the gut microbiome makes it a powerful tool for understanding and potentially influencing infant health and development.
