Multi-Omics Integrative Analysis of Serum and Sputum to Uncover Key Determinants of Prognosis in Patients With ARDS

This study is a prospective observational investigation designed to systematically characterize key microbial signatures, metabolite profiles, and gene-expression features in serum and sputum from patients with acute respiratory distress syndrome (ARDS), and to evaluate their associations with response to invasive mechanical ventilation and clinical outcomes. A total of 411 adult ARDS patients meeting the Berlin definition and receiving invasive mechanical ventilation will be enrolled; individuals with significant pre-existing pulmonary disease or a history of immunosuppression will be excluded. Blood and sputum specimens will undergo high-throughput sequencing and metabolomics, including 16S rRNA-based microbiome profiling, whole-transcriptome RNA sequencing, and targeted/untargeted metabolite quantification by LC-MS/MS. Ventilator-related parameters (e.g., tidal volume, positive end-expiratory pressure \[PEEP\], and respiratory system compliance) and clinical endpoints (e.g., 28-day mortality) will be integrated to establish a comprehensive multi-omics analytical framework. Data preprocessing will include batch-effect correction, normalization, and multiple-testing adjustment with false discovery rate control (FDR \< 0.05). Differential microbes, metabolites, and transcripts will be functionally interpreted using KEGG pathway and Gene Ontology (GO) enrichment analyses. Spearman correlation analyses will be performed to examine associations between omics features and ventilator parameters. Key features will be selected using LASSO regression, followed by development of random forest and support vector machine (SVM) models to predict mechanical ventilation response and the risk of ventilator-induced lung injury (VILI). Model performance will be assessed using ROC curves, area under the curve (AUC), calibration plots, and decision curve analysis. By integrating serum and sputum multi-omics data, this study aims to identify molecular biomarkers that influence the ef