Abstract

ObjectiveThis study aims to identify critical biomarkers and clarify how insomnia exacerbates sepsis-induced acute lung injury (SALI). We used integrative multi-omics approaches and machine learning.MethodsA causal association between sepsis and insomnia was established using Mendelian randomization (MR). We used weighted gene co-expression network analysis (WGCNA) to identify genes linked to both insomnia and SALI. We used machine learning techniques (Random Forest, SVM, KNN) with SHAP interpretability modeling to refine gene signatures. The diagnostic and prognostic value of these genes was investigated. To elucidate the underlying molecular pathways, functional enrichment analyses, including KEGG, GO, PPI, and GSEA were performed. To validate gene expression patterns and cellular localization, transcriptomic profiling, single-cell RNA sequencing (scRNA-seq), and in vivo and vitro experimental validation were employed.ResultsMR analysis identified insomnia as a causal determinant in susceptibility to sepsis. Complementary pathological evidence from preclinical sleep deprivation models further confirmed its role in exacerbating progression of SALI. The WGCNA revealed 1,294 co-dysregulated genes shared between insomnia and SALI. These genes were significantly enriched in biological processes, including immune regulation and phagocytic vesicle formation, as well as KEGG pathways such as tuberculosis infection and chemokine signaling. Among these,102 genes exhibited differential expression in a murine SALI model induced by LPS. Through machine learning analysis, ISG20, MYO1F, and PTPN6 were identified as robust hub genes. Further diagnostic stratification and prognostic evaluation prioritized PTPN6 as the most promising candidate. Immune infiltration analysis, scRNA-seq profiling and GSEA collectively demonstrated that PTPN6 expression is predominantly localized to macrophages and functionally involved in modulating the JAK/STAT3 signaling pathway. Functional validat