Abstract

Purpose Sepsis, a life-threatening condition with high mortality, is closely linked to energy metabolism (EM) and immune-inflammatory responses. However, the precise mechanisms remain incompletely understood. This research aims to identity EM-related genes (EMRGs) in sepsis and examine the diagnostic potential and molecular mechanisms through machine learning and single-cell RNA sequencing (scRNA-seq).Methods This study utilized the GSE65682 and GSE95233 datasets from the Gene Expression Omnibus (GEO) for analysis. Kaplan-Meier (KM) survival analysis and receiver operating characteristic (ROC) diagnostic analysis were applied to identify key EM-related genes. A sepsis model with overexpression of key genes was developed, and RNA sequencing (RNA-Seq) was employed to identify associated genes. Additionally, scRNA-seq was conducted to examine cell type distributions and gene expression profiles in sepsis.Results ATM and PEA15 were identified as critical genes. Overexpression of PEA15 alleviated septic symptoms. In sepsis, alterations in immune cell infiltration, particularly T follicular helper cells, were correlated with key gene expression. Core genes (RSAD2, IFI44, MX1, IFIT3, and ISG15), closely associated with the key genes, were also identified. Single-cell analysis further delineated the cell type profiles and core gene expression patterns in sepsis. From a translational perspective, PEA15 addresses a critical gap in sepsis management. Its strong prognostic and diagnostic performance, validated through KM and ROC analyses, positions it as a promising biomarker for early sepsis detection—essential for improving patient outcomes through timely intervention.Conclusion PEA15 and its associated core genes represent potential therapeutic targets: modulating PEA15 expression or targeting the underlying molecular network may help restore immune balance and reduce septic d