Abstract

Type 2 diabetes mellitus (T2DM) is increasingly linked with neuroinflammatory complications and progressive cognitive dysfunction; however, the molecular mechanisms underlying diabetes-associated neuronal impairment remain poorly understood. This study aimed to identify potential molecular biomarkers and predictive signatures associated with cognitive dysfunction in T2DM using integrated bioinformatic and machine learning approaches. Gene expression datasets related to T2DM and cognitive impairment were retrieved from the Gene Expression Omnibus (GEO) database, and common differentially expressed genes (DEGs) were identified followed by Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Protein–protein interaction (PPI) networks were constructed using the STRING database and visualized in Cytoscape software to identify central hub genes. Functional enrichment analysis revealed significant involvement of PI3K-Akt signaling, AGE-RAGE signaling in diabetic complications, inflammatory response pathways, oxidative stress regulation, and neuroactive ligand–receptor interaction pathways. Hub gene analysis identified TNF, IL6, AKT1, INS, VEGFA, and APP as major regulators associated with diabetic neurodegeneration. Two machine learning-based prediction models, Random Forest (RF) and Support Vector Machine (SVM), were developed using hub gene expression profiles, and both models effectively differentiated cognitively impaired diabetic samples from controls, while the RF model demonstrated superior predictive performance. Molecular docking studies further identified favorable interactions between selected phytocompounds and inflammatory target proteins. These findings suggest that integrated bioinformatic and machine learning approaches may provide valuable insights into biomarker discovery and therapeutic target identification for diabetes-associated cognitive dysfunction.