Research on screening key proteins related to the pathogenesis of sepsis associated encephalopathy based on DIA proteomics technology

Objective To explore the key proteins in plasma of patients with sepsis associated encephalopathy (SAE) through data-independent acquisition(DIA) quantitative proteomics and machine learning techniques, with the aim of guiding research on SAE pathogenesis and aiding early diagnosis. Methods The protein levels of plasma in patients with sepsis and SAE were measured using DIA quantitative proteomics technology. The differential proteins between the two patient groups were identified and functionally analyzed using bioinformatics methods. Key differential proteins were further screened using machine learning models such as Extreme Gradient Boosting （XGBoost） and Least Absolute Shrinkage and Selection Operator （LASSO） regression to construct a nomogram and evaluate its diagnostic performance. Results A total of 22 differentially expressed proteins were identified in the plasma of patients using DIA quantitative proteomics technology and bioinformatics analysis. Among these, the expression levels of 9 proteins were upregulated and 13 proteins were downregulated in the SAE group. These proteins were involved in signaling pathways such as Mitogen-Activated Protein Kinase （MAPK） signaling, focal adhesion and phospholipase D signaling pathways, and were involved in biological processes such as cell adhesion, signaling, macromolecular modification, etc. Meanwhile, intestinal diseases were found to have a potential association with the occurrence of SAE. Further screening was conducted using XGBoost and LASSO regression machine learning model, along with 5-fold cross validation to identify two key differential proteins, Endoplasmic Reticulum Protein 29 (ERP29) and Interleukin Enhancer Binding Factor 3 (ILF3), among the 22 differentially expressed proteins. A nomogram was constructed with these two key proteins and the diagnostic efficacy was evaluated, with area under curve (AUC) value 0.925 6 for the receiver operating characteristic curve (ROC) in the training set and AUC value 0.875 in the test set. Conclusion ERP29 and ILF3, identified through DIA quantitative proteomics and machine learning techniques, are key proteins involved in the occurrence of SAE and represent important targets for studying the pathogenesis of SAE. ERP29 and ILF3 can be used as biomarkers for early diagnosis of SAE.