Structural Analysis of a Multipartite Knowledge Network of Arrhythmia-Related Prescriptions in 『Donguibogam』

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Objectives: The rising burden of arrhythmia and limitations of current therapies highlight the need to systematically characterize the knowledge structure of traditional medicine. We conducted a multipartite network analysis of arrhythmia-related content in 『Donguibogam』 to quantitatively elucidate its knowledge organization.Methods: Arrhythmia-related texts were extracted from 『Donguibogam』 using predefined keywords derived from prior Korean literature. Texts were restructured into prescription-centered units; herbs, symptoms, and causes were identified through expert consensus-based tokenization and standardization. Unique herb combinations were analytical units, and binary presence matrices were constructed. An undirected, unweighted multipartite network of symptom-cause-herb relationships was generated from co-occurrence within prescriptions. Network and subnetwork analyses assessed structural properties, heterogeneity (Gini coefficient and entropy), and community structure using the Louvain algorithm.Results: A total of 225 unique herb combinations yielded a whole network comprising 68 symptoms, 25 causes, 232 herbs, and 3,464 edges. Low density (0.0658) and high heterogeneity (Gini 0.5591) indicated a hub-dominant structure. Subnetworks showed distinct patterns: the dizziness network exhibited the highest density and entropy with low inequality, whereas the consumptive disease and chest pain networks showed greater heterogeneity and hub dependence. Community analysis identified five major modules, with category-specific variation in internal density and degree inequality; subnetworks showed similar modular structures but distinct cohesion and heterogeneity profiles.Conclusions: This study reveals distinct structural patterns in arrhythmia-related knowledge within 『Donguibogam』 through network-based analysis. Despite methodological limitations, it provides an interpretable framework of knowledge organization and a foundation for future ontology-driven, clinically integrated decision-support models.