Identification of biomarkers for pediatric sepsis based on machine learning and bioinformatics analysis

Pediatric sepsis is a systemic inflammatory syndrome caused by dysregulated host immune responses, with a high mortality rate and a lack of effective biomarkers, posing significant challenges for early diagnosis and treatment. This study integrated six bulk RNA-seq datasets related to pediatric sepsis, including 497 patients and 116 healthy control samples. Weighted gene co-expression network analysis was used to identify gene modules significantly associated with pediatric sepsis, and 237 high-confidence biomarkers were screened based on 14 machine learning models, among which RORA and GPR183 stood out in multiple models. Functional analysis indicated that these biomarkers were mainly involved in biological processes such as transcription and translation, the immune system, and cellular senescence. Immune infiltration analysis revealed a significant reduction in adaptive immune cells such as B cells and CD8 + T cells and an increase in neutrophil and monocyte infiltration in pediatric sepsis patients, consistent with the “immunoparalysis” theory. Notably, RORA and GPR183 were positively correlated with CD8 + T cells, suggesting their potential role in regulating T cell function. Additionally, we developed an open-source website for real-time application of biomarkers. Furthermore, we established a sepsis model in zebrafish and found that rora1 , rora2 , and gpr183 expression levels were significantly downregulated in the disease group. This study provides new insights for developing novel diagnostic tools and targeted therapies for pediatric sepsis.