Date of Award

Fall 9-8-2025

Degree Type

Thesis

Degree Name

Master of Science (MS)

Department

Bioinformatics & Computational Biology

First Advisor

Catherine Putonti

Abstract

Many infections can be attributed to a specific microorganism. For example, while there can be many causes of urinary tract infections (UTIs), such as the bacterial species Klebsiella pneumoniae or Enterococcus sp., the most widespread cause for UTIs is uropathogenic Escherichia coli. Generally, diseases or infections that are the result of a single pathogenic organism are more easily understood. Other diseases, such as irritable bowel disease (IBD), cannot be attributed to one member in the community; rather, it is a result of dysbiosis of the microbial community (microbiota) of the gastrointestinal tract, a combination of an overgrowth of problematic members and a reduction of beneficial members. As such, the dynamic nature of such communities and their contributions to diseases is yet to be fully understood. The recent advancement of computational tools has enabled us to explore the roots of diseases due to dysbiosis. More specifically, scientists are developing predictive models to understand how biomarkers and microbial community compositions are associated with different disease states. Collectively, these studies serve as a precursor for future studies involving artificial intelligence applications to the microbiome. However, we need far more robust and extensive amounts of data for application to healthcare and personalized medicine. Further research will allow us to explore the novel, underlying causes of infections due to dysbiosis and understand the scope of metadata needed to improve the predictive power of these algorithms, in turn improving our ability to better diagnose and treat them in the future. This thesis work utilizes interpretable machine learning algorithms to identify potential sources of infection, such as opportunistic pathogens, genetic content, and taxa contributing to dysbiosis.

Download

Share

COinS