This study pioneers the systematic identification of core genes associated with mitochondrial dysfunction in pulmonary arterial hypertension (PAH) through multidimensional bioinformatics integration and machine learning algorithms. By elucidating their expression profiles, diagnostic efficacy, and mechanisms of association with immune infiltration, it provides new evidence for molecular mechanism research and clinical translation of PAH. Using GEO database resources, we obtained the training set GSE38267 (13 PAH cases + 28 healthy controls) and validation set GSE15197 (18 PAH samples + 13 normal controls). Through differential expression analysis and mitochondrial gene set intersection screening, we identified differentially expressed genes related to PAH-mitochondrial dysfunction. Subsequent analyses including GO/KEGG enrichment, PPI network construction, Cytoscape core node screening, combined with three machine learning algorithms (LASSO regression, SVM-RFE, and RF), confirmed core genes via validation set. Diagnostic value was assessed using ROC curves, while ssGSEA, Cibersort, and correlation analyses revealed associations between core genes and immune infiltration. Results demonstrated that ISCA1, STOM, NT5M, ACSL6, and ALAS2 are core genes associated with mitochondrial dysfunction in PAH. Their expression abnormalities contribute to disease pathogenesis by regulating mitochondrial energy metabolism, structural homeostasis, and immune cell infiltration. Notably, the area under the ROC curve (AUC) for ISCA1, STOM, and NT5M was significantly higher than 0.9, demonstrating outstanding potential for PAH diagnosis.