EVALUATION OF THE EFFECTIVENESS OF APPLYING AN INFORMATION HYBRID SYSTEM FOR THE CLASSIFICATION OF CELLULAR ATYPIA IN PAP TESTS; Hodnotenie efektívnosti aplikácie informačného hybridného systému pre klasifikáciu bunkovej atypie v PAP testoch

Relevance. In modern healthcare, cervical cancer screening with Pap tests is a cornerstone of efforts for early detection of this disease in the female population. However, the interpretation of Pap test results applying traditional microscopy may be subject to variability and subjective assessment among different specialists, which requires the development of a reliable analytical framework. The introduction of digital microscopy into medical diagnostics will allow to objectify the diagnostic process and the effectiveness of detection and classification of cellular changes. Objective. To evaluate the effectiveness of application of the developed architecture of the hybrid information system in diagnostics of cellular atypia in Pap tests. Methods. The mechanism for developing a prototype of the information hybrid system for describing cellular transformations in Pap tests included the following steps: study of the subject area, construction of the conceptual apparatus of the system; study of parameters for objectification, determination of value ranges by analyzing the world literature; selection of a software platform for implementation, testing of the system together with medical experts by means of digital shooting of clinical material and comparison of hypotheses issued by the system and the opinion of medical staff with the compilation of a matrix of adjacency of hypotheses of cellular transformations according to the Bethesda classification from 2014 and calculation of the following parameters of the accuracy of the applied system: accuracy (Acc, Accuracy), specificity (Spec, Specifity), sensitivity (Sens, Sensitivity) and area under the curve (ROC AUC). Results. Based on the conducted analysis, a prototype of the information hybrid system "Information hybrid system for describing cellular transformations of Pap tests" was developed, which has two basic blocks: 1) "Objectifier of images with Pap tests cells" ‒ the section contains a program developed taking into account the structure of the knowledge base that automatically processes images and extracts objective parameters from them; 2) "Knowledge base for objective description of cellular changes and its Solver" ‒ the module contains objective indicators determined during the study of global literature with the corresponding input fields for users, characterizing with the help of the Solver hypotheses of assignment to a particular type of cellular transformation. Together with cytologists of the FEFU Medical Complex, an assessment was made of the coincidence of reasoning about the belonging of cells to a particular class of atypia of the developed system with the opinion of experts on 100 images. The following parameters of the accuracy of the developed information architecture of the system were obtained: Accuracy = 78%, Sensitivity = 79% and Specifity = 89% with AUC = 83%. Conclusion. The proposed prototype of the information hybrid system demonstrates the possibility of objectification of cellular transformations observed in Pap tests. By combining computational methods with expert knowledge, the system allows for a standardized approach to cytological analysis, thereby potentially increasing the accuracy and effectiveness of cervical lesion screening programs. © 2025 Elsevier B.V., All rights reserved.