Objective To identify anoikis-related genes and immune infiltration characteristics in pancreas islet tissues involved in the pathogenesis of type 2 diabetes mellitus (T2DM) using bioinformatic analysis. Methods The dataset GSE76894 was downloaded from the Gene Expression Omnibus (GEO) database as the training set. Diffe-rential gene expression analysis was conducted on T2DM and non-diabetic islet tissues within the training set, and intersected with the anoikis-related gene set to obtain anoikis-related differentially expressed genes (DEGs). Subsequently, key genes were identified using the random forest (RF) and least absolute shrinkage and selection operator (LASSO) algorithms. Receiver operating characteristic (ROC) curves were plotted, and the area under the curve (AUC) was calculated to evaluate the association strength between the expression levels of the identified key genes in pancreas islet tissues and T2DM, followed by validation in the GSE76895 dataset. Protein-protein interaction (PPI) network analysis and gene ontology (GO) enrichment analysis were then performed on the identified key genes. The immune infiltration analysis was conducted using the CIBERSORT algorithm. Results Differential analysis identified 8 anoikis-related DEGs, with 6 upregulated and 2 downregulated genes. Subsequent application of two machine lear-ning algorithms identified 4 key genes: PDK4, BMF, ITGB1, and SNAI2. ROC analysis showed that in the validation set (GSE76895), only PDK4 expression had strong discriminatory power (AUC = 0.721), indicating a significant association with T2DM. Enrichment analysis demonstrated that these key genes were primarily enriched in terms related to integrin-mediated cell adhesion, regulation of lipid biosynthetic processes, integrin complex, and glial cell protrusions. Immune infiltration analysis indicated differential expression of various immune cells in the pancreas islet tissues of T2DM and healthy individuals. Furthermore, PDK4 expression was negatively correlated with that of M0 macrophages. Conclusions PDK4 is downregulated in T2DM islet tissues and negatively correlated with M0 macrophage expression levels, suggesting that the expression of PDK4 is related to the T2DM pathogenesis caused by immune dysregulation to some extent.