Hereditary diseases, which are as orphan (rare) pathology, have more than 6,000 different phenotypes. For patients with certain metabolic diseases characterized by a progressive course, early determination of an accurate diagnosis is the basis for preventing the development of the disease with the timely appointment of treatment with effective drugs. The result is an increase in the quality of life, and in some clinical forms, a complete cure for the disease. However, the differential diagnosis of these genetic diseases is associated with several difficulties due to the extremely low frequency of occurrence in the population and the similarity of phenotypic manifestations. At the same time, an error at the stage of clinical examination can complicate and delay the necessary confirmation of the diagnosis at the stage of laboratory or molecular genetic testing. The research purpose is to improve the accuracy and timeliness of diagnosis in hereditary lysosomal storage diseases at the pre-laboratory stage of differential diagnosis using a computer expert decision support system. Materials and methods. The extraction of knowledge about orphan diseases was a two-stage process. At the first stage, the research material was Russian and foreign sources of information on the clinical manifestations of hereditary lysosomal storage diseases - monographs, guidelines, and articles, including descriptions of clinical cases, as well as online information databases. In the process of knowledge extraction, semantic, textological and linguistic methods of text analysis were used. For the structured presentation of knowledge, a specially developed textological card was used. At the second stage, the knowledge recorded in the textological cards was refined by experts, who assessed all signs using confidence measures. Results and its discussion. The structure of the system has been developed, which includes a user interface, a knowledge base, a database (work area), and a block of explanations. The knowledge base is implemented based on an integral model of the disease, which includes an expert assessment of modality and confidence factors of signs, as well as taking into account the age-related characteristics of the manifestation of the diagnosed pathology. The comparative analysis algorithm provides a comparison of a specific case with the reference version of the integral model and provides the possibility of ranking the diagnostic hypotheses put forward by the system. The explanation block allows the user to present information about the signs that support the hypotheses that are missing for unambiguous confirmation of the hypothesis and are not related to the proposed diagnosis. The system is implemented on the IACPaaS ontological platform. The test results showed a high (over 80%) efficiency of identification of nosological forms in two samples: (1) clinical cases with a verified diagnosis based on data from Russian and foreign publications, (2) a sample of verified data from case histories of a specialized Russian clinic. Conclusion. Expert diagnostic decision support system for lysosomal hereditary diseases with an expandable knowledge base has been developed.
Download Journal of New Medical Technologies №4 PDF from the Medical Institute TSU archive: http://medtsu.tula.ru/VNMT/Bulletin/2021/21B4.pdf
Blagosklonov N. A., Kobrinskii B. A. Ekspertnaya sistema dlya diagnostiki nasledstvennykh zabolevaniy [Expert system for diagnostics of
hereditary diseases]. Journal of New Medical Technologies. 2021;4:98-102.