original_data

Background: Blood gas parameters and immunity are associated with the prognosis of septic patients. The aim of this work is to develop a blood gas parameters-based signature (BGBS) that could be used to predict patient prognosis and immune responses.Methods: Medical records retrospectively extracted from the Medical Information Mart for Intensive Care IV were used for constructing BGBS in septic patients and for internal validation in septic shock patients. Bioinformatics analysis and machine learning techniques were adopted to reveal the relationship between BGBS and the primary outcome, 28-day mortality, in sepsis and septic shock.Results: A BGBS containing PO2, BE, and lactate was developed and verified as an independent and reliable risk model that correlated with the 28-day mortality of septic patients (odds ratio: 1.559; 95% confidence interval: 1.464-1.659; P <.001). Moreover, its prognostic performance was internally validated in septic shock patients (odds ratio: 1.432; 95% confidence interval: 1.321-1.552; P <.001). More severe sepsis was observed in high-risk patients than in low-risk patients as determined by BGBS. Subsequent investigations demonstrated a positive relationship between the BGBS-calculated risk score and immunosuppression, suggesting that innate immunity might mediate the predictive role of our identified signature in 28-day mortality of septic patients. In addition, a BGBS-derived nomogram was constructed and demonstrated to be superior to existing prognostic scales.Conclusion: We developed and validated a novel signature and nomogram that could rapidly evaluate the prognosis of septic patients and identified a subset potentially suitable for immunoenhancement therapy and intensified treatment.