Abstract
Sepsis is a life-threatening condition caused by a dysregulated host response to infection and is one of the leading causes of morbidity and mortality worldwide. Glycosylation is one of the key modes of protein modification, affecting protein folding, transportation, and localization. Glycosylation patterns are closely related to sepsis, but their specific impact still needs further investigation. This study explored the role of glycosylation-related genes in sepsis through bioinformatics analysis and machine learning, and validated the expression value of the key genes. We identified 38 differentially expressed glycosylation-related genes in sepsis datasets, which divided sepsis patients into two subgroups with different survival outcomes, thus highlighting their prognostic value. Subsequently, we constructed prognostic models using various machine learning methods, classifying patients into high-risk and low-risk groups with significantly different survival rates. We conducted biological analysis of the key genes in the model at the single-cell level and also validated the expression of these key genes in sepsis patient samples. Our study not only enhances the understanding of sepsis glycosylation but also provides a new strategy for clinical diagnosis and prognosis.