Jack Gryebet
Research Mentor: Jean Nemzek
Mentor Department: Unit for Laboratory Animal Medicine, Medicine
Author(s): Jean Nemzek, Oliver He
Session: Session 2 (10:00 AM – 10:50 AM)
Presentation Type: Poster 118
Abstract
Introduction: Experimental sepsis studies using cecal ligation and puncture (CLP) models have reported conflicting effects of Tumor Necrosis Factor (TNF), a pro-inflammatory cytokine, in sepsis. Some suggest beneficial while others suggest harmful roles, and despite promising TNF-targeted findings in animal models, new strategies haven’t translated into effective treatment of sepsis in humans. This project aimed to create an ontology to classify and annotate the methods and outcomes from preclinical sepsis studies. The ontology would then be used to inform a large language model and develop an AI tool capable of in depth meta-analysis and other functions. Method: PubMed was the database searched for of all analyzed studies. The search criteria included all CLP animal studies investigating sepsis-related mortality in relation to TNF that were written from 2000 to 2026. Ineligible studies: did not track mortality rate, were other meta-analyses, or were inaccessible. CLP studies were evaluated to create an ontology mapping variables affecting experimental outcomes. With AI assistance, we compared individual and grouped variables and analyzed how they affected the mortality rate outcome. Results: A total of 30 papers were included in the analysis and over 220 experimental groups were analyzed. There were 23 variables identified specific to mouse, environment, and intervention that may influence the CLP model and outcomes. These variables were used to create an ontology that informed a large language model and allowed development of an AI tool capable of analyzing data gleaned from previous studies. Conclusions: Lack of model standardization and incomplete reporting of methods in published literature make comparisons of studies difficult. An ontology was developed using key variables that influence experimental outcomes in cecal ligation and puncture models. By identifying patterns between variables and outcomes, we will provide a guide for researchers to plan future studies with more ideal and translatable experimental conditions. An AI tool will also help better define the role of mediators such as TNF-alpha in sepsis through faster and unbiased evaluation of preclinical studies.
