A Machine Learning-Based Diagnostic Nomogram for Moyamoya Disease: The Validation of Hypoxia-Immune Gene Signatures

Moyamoya disease (MMD) is a cerebrovascular disease which can result in severe strokes. However, its etiology is still unknown. We analyzed gene expression datasets from 36 MMD patients and 24 controls to identify differentially expressed genes. Using weighted gene co-expression network analysis and databases such as KEGG, we identified hypoxia-immune-related genes. These genes were further refined through machine learning algorithms. The diagnostic value was confirmed using an external dataset, and a diagnostic nomogram was constructed. Additionally, gene set enrichment analysis was conducted, and a competitive endogenous RNA (ceRNA) network was built to predict potential therapeutic targets. Our study identified AKT1, CLDN3, ISG20, and TGFB2 as the key hypoxia-immune genes associated with MMD. These genes were implicated in epithelial-mesenchymal transition, angiogenesis, and cell adhesion, suggesting a role in MMD pathogenesis. Further, our study constructed the ceRNA network and predicted potential drug targets for MMD. We obtained the top 10 drugs in the interaction of the four key genes that might serve as potential targets for the treatment of MMD. In conclusion, this study comprehensively analyzes the role of hypoxia-immune genes in MMD, which is conducive to the development of new diagnostic and therapeutic approaches and the exploration of the potential pathogenesis of MMD.