Knowledge graph construction and latent knowledge mining based on the Dan-an CHENG’s Acupuncture and Moxibustion Experience Collection

Objective This study employed knowledge graph (KG) construction and graph data mining techniques to systematically elucidate Dan-an CHENG’s academic experience in acupuncture diagnosis and treatment as well as explore latent therapeutic patterns and inherent correlations embedded in his medical records, to provide a reference for the quantitative analysis and digital preservation of acupuncture case literature. Methods: Using 467 medical records from the Dan-an CHENG Acupuncture Experience Collection as the study dataset, clinical entities and semantic relationships were extracted from the texts. Following the structured “Diagnosis–Prescription–Technique” framework, the Dan-an CHENG Knowledge Graph (DaCKG) was constructed, with the Neo4j graph database supporting the storage, retrieval, and visualization of acupuncture diagnostic and therapeutic knowledge. Graph algorithms from the Graph Data Science (GDS) library were applied to analyze the associations among diseases, syndromes, and core needling and moxibustion techniques (NMT). The analysis incorporated treatment intervention timing, stimulation intensity, needle insertion depth, stimulation duration, and reinforcing and reducing manipulations to identify latent therapeutic patterns. Results The DaCKG integrated the diagnostic, prescription, and technical dimensions of acupuncture. The ontology schema defined 10 entity categories and 23 semantic relationships, and the final KG comprised 4,843 nodes and 20,987 relationships, enabling multidimensional correlation retrieval and visualization across diseases, syndromes, acupoints, and manipulations. Analysis using GDS algorithms demonstrated that Dan-an CHENG’s acupuncture prescriptions do not conform to fixed paradigms, but form a dynamic treatment system guided by syndrome differentiation, in which different NMT is applied to different acupoints within a single prescription, and varied NMT regimens are adopted for the same acupoint when treating different diseases and syndromes; importantly, reinforcing and reducing manipulations act as the stable core technique across prescriptions, while parameters including needle insertion depth, stimulation intensity, and treatment intervention time are dynamically adjusted in line with syndrome types and disease stages. Conclusion Dan-an CHENG’s acupuncture prescriptions use syndrome differentiation as the fundamental guideline, adhering to stable core therapeutic principles while allowing flexible clinical adjustments. His NMT was adjusted according to syndrome patterns, and his prescription formulations exhibited both regularity and individualized variation. The DaCKG transformed unstructured medical records into structured diagnostic and therapeutic knowledge. It provides a computable platform for the digital preservation of Dan-an CHENG’s academic experience but also offers a methodological reference for the collation and knowledge mining of medical records from renowned acupuncture experts.