Identification of pyroptosis-related molecular clusters in alzheimer’s disease based on multiple machine learning models

Abstract

AbstractBackground:Alzheimer's disease (AD) is the most common neurodegenerative disease. Pyroptosis is a new type of programmed cell death, which can lead to the progression of various diseases. The aim of this study was to explore the role of pyroptosis-related genes (PRGs) in Alzheimer's disease and to build the predictive model.Methods:The expression of PRGs in AD was analyzed based on the GSE33000 dataset, and molecular clustering and immune microenvironment analysis were performed on 310 patient samples. The WGCNA algorithm was used to identify the genes that were specifically expressed between different clusters, and then four machine learning models (RF, GLM, SVM and XGB) were used to construct the predictive models for the risk of AD. The prediction capability of the model was verified by nomogram, calibration, decision curve analyses and five external data sets.Results:Multiple PRGs were differentially expressed between AD and normal brain tissue. Based on differentially expressed PRGs, 310 AD patients were divided into two subtypes by consistent clustering. Immune microenvironment analysis showed significant differences in the degree of immune activation among different subtypes. WGCNA algorithm identified the specific genes between AD and normal individuals, Cluster 1 and Cluster 2. The SVM model has the best prediction performance with low residual error and root mean square error, and high area under ROC curve (AUC=0.933). Finally, a prediction model based on five genes (GPR4, STAT3, CASP4, CLIC1 and TNFRSF10B) was constructed and showed satisfactory performance on five externally validated data sets. Nomogram, calibration curve and decision curve analysis proved the prediction performance of the model.Conclusions:This study systematically analyzed the complex relationship between PRGs and AD, and constructed a good prediction model to distinguish AD from normal individuals, which is expected to provide reference for related research.