Construction Of Knowledge Graph Of Physicians’ Treatment Based On Syndrome Differentiation For Patients With Insomnia

Objective:To abstract the framework of knowledge ontology of different Traditional Chinese Medicine(TCM)physicians’ Treatment based on Syndrome Differentiation(TBSD);Explore the construction process of knowledge graph(KG)of TBSD based on real world observational research data;visualize different TCM physicians’ TBSD knowledge system;provide new tools and systematic procedure for mining and analyzing the characteristics of physicians’ latent thinking,diagnosis and treatment; lay methodological foundation for the inheritance and innovation of TCM physicians,experience.Methods:1)Data collection and pre-processing of real process of TBSD for patients with insomnia:Based on the key project sponsored by the national natural science foundation of china(stepwise progressive evaluation method of TBSD)and real world clinical process,we collected the clinical data of different TCM physicians prospectively and continuously for a period of time by the TCM clinical research information sharing system.Medical Integrator Studio was taken as the pre-processing software.Clinical data of different TCM physicians was pre-processed through the data extraction,transformation and loading.2)The construction of the framework of the ontology of TBSD based on literature research:CNKI,CBM,WanFang and VIP databases were systematically searched using the terms "TCM","medical record","case report",and "guideline","standard",and "clinical pathway",and relevant teaching materials including TBSD were searched manually.We take NoteExpress as the Literature management software.Qualified literatures were distinguished by initial and full-text screening,and the basic elements of TBSD were extracted.We use Gephi as the software for co-occurrence analysis and visual display.3)Construction of KG combined with the optimization strategies of complex networks:we construct the preliminary KG of different TCM physicians mainly based on the real world clinical data,through the steps of data export,co-occurrence matrix construction,data annotation,data import,data verification and KG construction.Then we modify and optimize the preliminary KG in a comprehensive way by means of expert interview,literature and complex networks The knowledge map is constructed by Gephi software,and the complex network is constructed by Liquorice software.Results:1)Clinical database of Three TCM physicians:As of December 31,2016,clinical data of 431 cases and 1022 visits were included in the three cohorts:85 cases and 259 visits for physician A,101 cases and 245 visits for physician B;245 cases and 518 visits for physician C.A individualized pre-processing strategy of "human-machine integration and people-oriented" has been formed,including steps of data retrieval,verification,preparation of pre-processing rules,rule feedback and revision,rule import and batch processing,post-processing verification.We pre-processed the data of symptoms,diagnosis,therapeutic principles and methods,formula,Chinese medicine,western medicine/Chinese patent medicine,laboratory examination,and physical examination according to 11119 rules.2)Framework of the TBSD ontology:5555 literatures were retrieved,1912 repeated literatures were deleted,370 literatures were obtained through title and summary screening,328 literatures were deleted through full-text screening,and 42 literatures were finally included,involving 22 guidelines,17standards,1regulation and 2 sets of teaching materials.Framework of the TBSD ontology was constructed,including four domains of" clinical manifestations"," diagnosis"," treatment"," clinical evaluation",53 basic elements and 9 types of connections.3)Construction and optimization of KG:We constructed the clinical manifestations,diagnosis,and treatment KG of 3 different TCM physicians.Combined with methods of complex networks,literature research,and expert interview,the core formula of the three TCM physicians and their indications were excavated and the optimized KG of TBSD were formed based on the core indication population.A set of systematical procedures for constructing TCM KG for TCM physicians was initially formed.Conclusion:We took three TCM physicians as an example to build the individualized KG TBSD,and initially formed a set of procedures for constructing KG of different TCM physicians TBSD process.Results shows that KG based on real-world clinical data is an effective technology to visualize the individualized different physicians’ TBSD process of insomnia.TCM clinical research information sharing system could provide structured,homologous real-world clinical data,and lays a good data foundation for the construction of TCM physician’s KG.Through steps of the "human-machine integration and people-oriented" data pre-processing,construction of TBSD ontology,mapping of TBSD ontology and clinical data,construction and optimization of KG,clinical evaluation,and the excavation of core formula,we could build up the individualized KG of different physicians’ TBSD process of insomnia.The individual KG combined with the clinical evaluation of TBSD could provide new tools and systematical procedures for mining and visualize TCM physicians’ latent thinking,diagnosis and treatment characteristics,laying a mythological foundation for the inheritance and innovation of TCM physicians experience.