Научно-практический журнал Медицинские технологии. Оценка и выбор
  • Русский
  • English

Development of Knowledge Base Architecture for Clinical Decision Support System Based on Graph Database

Issue № 33 | 2018 (3)

DOI: https://doi.org/10.31556/2219-0678.2018.33.3.042-048 meta


High urgency of healthcare informatization and features of the organization of medical and diagnostic process allow searching for the ways of maintenance of medical activity information support. One of such ways is the development of clinical decision support systems based on knowledge. Among the forms of knowledge representation, a semantic network is distinguished, structurally repeating the model of graph databases that have the necessary advantages for working with knowledge.

Purpose of the study. Development of the knowledge base architecture of the clinical decision support system for the instrumental diagnosis of angina pectoris based on the ontological approach, using a graph database.

Materials and methods. Extraction of concepts related to the diagnosis of angina pectoris was carried out from clinical recommendations concerning stable ischemic heart disease. The primary accumulation of knowledge and grouping of concepts by types was carried out in MS Excel; the elaboration of the types of connections between the highlighted concepts was carried out in the ARIS Express program. To form the knowledge base, a Neo4j graph database was used.

Results. From the clinical recommendations with the help of cardiologists, 401 root concepts and 619 synonyms were extracted, which were grouped by type of diagnosis, synonym, clinical sign, diagnostic test, diagnostic sign, laboratory test, medical personnel, location of the study. The groups of concepts were linked by means of links: parent-child, synonym, concomitant pathology, clinical sign, diagnostic sign, indication for the study, place of the study, person conducting the study, person interpreting the results of the study, person taking biomaterial sampling. To indicate the characteristic values, the graph database was used to fill the attributes of nodes and links, which made it possible to reduce the dimension of the graph. The created knowledge base was twice validated for the completeness and adequacy of solutions offered to the doctor using depersonalized electronic medical records of patients. The first validation returned incomplete compliance with appointments from electronic medical records, which made it necessary to refine the filling of the database with new knowledge. New knowledge was added to the database without the need to modify its architecture, after which the re-validation returned a complete match between the proposed and the actual assignments.

Conclusion. The use of clinical decision support systems based on the ontological approach using graph databases can be promising in terms of providing speed and explaining the proposed assignments. Properly organized architecture allows you to scale the knowledge base, and graph database features allow you to reduce the dimensionality of the graph, simplifying the work with knowledge.


angina pectoris, clinical decision support systems, CDSS, ontology, clinical recommendations.

For citations

Kiselev K. V., Noeva E. A., Vyborov O. N., Zorin A. V., Potekhina A. V., Osyaeva M. K., Shvyrev S. L., Martynyuk T. V., Chazova I. E., Zarubina T. V. Development of Knowledge Base Architecture for Clinical Decision Support System Based on Graph Database. Medical Technologies. Assessment and Choice. 2018; 3(33): 42–48.