Construction of Diagnostic Model for Regulatory T Cell-Related Genes in Sepsis Based on Machine Learning

Background: Sepsis is a complex syndrome caused by a severe infection that occurs with a severe inflammatory response. Regulatory T cells (Tregs) have immunosuppressive effects and play a crucial role in modulating the immune response. There-fore, the number of Tregs is significantly increased in sepsis patients. Methods and Results: This paper aims to identify Tregs associated with the diagnosis of sepsis. For this purpose, transcriptional data from the GEO database for sepsis and its controls were downloaded and subjected to differential expression analysis. Immuno-infiltration analysis of the obtained DEGs revealed that Tregs were significantly different in sepsis and its controls. To further explore the cellular landscape and interactions in sepsis, single-cell RNA sequencing (scRNA-seq) data were analyzed. We identified key cell types and their interactions, including Tregs, using cell-cell communication analysis tools such as CellChat. This analysis provided in-sights into the dynamic changes in immune cell populations and their communication networks in sepsis. Thus, we utilized multiple machine learning algorithms to screen and extract Treg-related genes associated with sepsis diagnosis. We then performed both in-ternal and external validation tests. The final diagnostic model was constructed with high diagnostic accuracy (accuracy of 0.9615). Furthermore, we verified the diagnostic gene via a qPCR experiment. Conclusions: This paper elucidates the potential diagnostic targets associated with Tregs in sepsis progression and provides comprehensive understanding of the immune cell interactions in sepsis through scRNA-seq analysis.