Application of machine learning reveals diagnostic biomarkers related to pyroptosis in Alzheimer's disease and analysis of immune infiltration

Background Alzheimer's disease (AD) is characterized by complex pathological mechanisms, with pyroptosis potentially contributing to neuroinflammation. Objective To identify pyroptosis-related genes (PRGs) in AD and explore their role in neuroinflammation, aiming to provide potential biomarkers and therapeutic targets for precision medicine in AD treatment. Methods Transcriptomic data from AD brain tissues (GEO database) were analyzed using multi-omics integration and machine learning. Key PRGs were screened via weighted gene co-expression network analysis (WGCNA), LASSO regression, random forest, and SVM-RFE algorithms. Molecular subtypes and therapeutic potential were assessed through unsupervised clustering and molecular docking. Results Analysis identified 609 differentially expressed genes (DEGs), with upregulated genes enriched in DNA transcription and mitosis-related pathways. Six core PRGs (MIB1, TUG1, GATA1, CA1, CFH, IL17A) demonstrated strong diagnostic accuracy (AUC > 0.85). Unsupervised clustering revealed two AD subtypes: a high-risk subtype with activated pyroptosis-inflammatory pathways and distinct immune microenvironment features (p < 0.05). Molecular docking confirmed stable binding between CFH and the anti-AD drug candidate fludrocortisone (binding energy < −7 kcal/mol). Conclusions Pyroptosis modulates neuroinflammation to drive AD progression. CFH and other PRGs serve as promising biomarkers and therapeutic targets, advancing precision strategies for AD subtyping and intervention.