Identification of novel inflammatory response-related biomarkers in patients with ischemic stroke based on WGCNA and machine learning

Background Ischemic stroke (IS) is one of the most common causes of disability in adults worldwide. This study aimed to identify key genes related to the inflammatory response to provide insights into the mechanisms and management of IS. Methods Transcriptomic data for IS were downloaded from the Gene Expression Omnibus (GEO) database. Weighted gene co-expression network analysis (WGCNA) and differential expression analysis were used to identify inflammation-related genes (IRGs) associated with IS. Hub IRGs were screened using Lasso, SVM-RFE, and random forest algorithms, and a nomogram diagnostic model was constructed. The diagnostic performance of the model was assessed using receiver operating characteristic (ROC) curves and calibration plots. Additionally, immune cell infiltration and potential small molecule drugs targeting IRGs were analyzed. Results Nine differentially expressed IRGs were identified in IS, including NMUR1, AHR, CD68, OSM, CDKN1A, RGS1, BTG2, ATP2C1, and TLR3. Machine learning algorithms selected three hub IRGs (AHR, OSM, and NMUR1). A diagnostic model based on these three genes showed excellent diagnostic performance for IS, with an area under the curve (AUC) greater than 0.9 in both the training and validation sets. Immune infiltration analysis revealed higher levels of neutrophils and activated CD4 + T cells, and lower levels of CD8 + T cells, activated NK cells, and naive B cells in IS patients. The hub IRGs exhibited significant correlations with immune cell infiltration. Furthermore, small molecule drugs targeting hub IRGs were identified, including chrysin, piperine, genistein, and resveratrol, which have potential therapeutic effects for IS. Conclusion This study confirms the significant impact of IRGs on the progression of IS and provides new diagnostic and therapeutic targets for personalized treatment of IS.