CONCEPT FOR A CLINICAL DECISION SUPPORT SYSTEM
IN GASTROINTESTINAL ENDOSCOPY 

Vladimir V. Khryashchev, Candidate of technical sciences, associate professor, associate professor of the sub-department of digital technologies and machine learning, P.G. Demidov Yaroslavl State University (14 Sovetskaya street, Yaroslavl, Russia), v.khryashchev@uniyar.ac.ru

Background. The article considers the problem of constructing a system for supporting medical decisionmaking for endoscopic examinations of the gastrointestinal tract based on neural network methods and video data analysis algorithms. To solve this problem, the structure of an artificial intelligence system is proposed that takes into account the specifics of video endoscopic examinations. Materials and methods. Deep machine learning methods are used to construct the system. Neural network architectures of the YOLO family, as well as transform architectures, are used as the core for key algorithms for processing the video stream. Results and conclusions. The concept of constructing a systemfor supporting medical decision-making is proposed. A trial operation of the system was carried out in two medical institutions. Studies have shown that the system can be used both directly during an endoscopic examination and for quality control of a screening study after it has been conducted and for recording the results in a medical information system.

decision support system, endoscopy, deep machine learning, clinical test, screening test

Khryashchev V.V. Concept for a clinical decision support system in gastrointestinal endoscopy. Izmerenie. Monitoring. Upravlenie. Kontrol' = Measuring. Monitoring. Management. Control. 2025;(2):137–144. (In Russ.). doi: 10.21685/2307-5538-2025-2-16