Multiscale Interactome-Guided Discovery Candidate Herbs and Active Ingredients Against Hyperthyroidism by Biased Random Walk Algorithm

Hyperthyroidism features excess thyroid hormone and a hypermetabolic state; although drugs and definitive therapies exist, mechanism-anchored options are still needed. We built a multiscale interactome and applied a biased random-walk diffusion model to prioritize herbal candidates, active ingredients, and mechanisms. Herb-compound records came from OASIS; targets from DrugBank, TTD, and STITCH; and disease genes from DisGeNET. For each herb and compound, we simulated diffusion profiles, computed the correlation with the hyperthyroidism profile, and assessed target overlap ratio. Herbs were ranked by correlation and <i>p</i> < 0.05 overlap, retaining those with ≥5 active compounds linked to disease targets. Top signals included <i>Geranii Herba</i> (0.021), Gastrodiae Rhizoma (0.012), and <i>Veratri Rhizoma Et Radix</i> (0.011), plus seven herbs at 0.010. Herb-disease relationships were strongly enriched. Enrichment analyses highlighted MAPK, PI3K-AKT, p53, HIF-1, and thyroid hormone signaling, with Gene Ontology terms for apoptosis/anoikis, inflammation, and RNA polymerase II-dependent transcription. Compound-level analysis recovered evidence-supported ellagic acid and diosgenin and proposed resveratrol, cardamomin, 20-hydroxyecdysone, and (Z)-anethole as novel candidates. Subnetwork mapping linked these compounds to phosphorylation, GPCR-cAMP/TSH signaling, and transcriptional control. This framework recapitulates known thyroid-modulating herbs and elevates underappreciated leads with testable mechanisms, supporting the discovery of multi-target therapeutics for hyperthyroidism.