Dovgich S. Prediction of Chronic Rhinosinusitis Recurrence Risk Using Modern Information Technologies (Neural Networks)

The dissertation presents the results of a clinical retrospective study based on the analysis of data from patients with chronic rhinosinusitis, which is widely prevalent in the population. According to diagnoses established solely on the basis of patient complaints, the prevalence of CRS varies from 10% to 28% in different countries, with an additional 4–9% of the population where the diagnosis is established "incidentally" after additional examinations such as endoscopy or computer tomography of the paranasal sinuses [1-5]. According to statistics from the UK and Wales, the frequency of outpatient visits for CRS to family doctors and otolaryngologists is approximately 15% [6]. Chronic rhinosinusitis is a significant medical and social problem as it leads to a considerable deterioration in patients' quality of life and burdens the healthcare system and the country's economy [7]. Predicting the risk of chronic rhinosinusitis recurrence is an important aspect of modern medical practice. With the introduction of modern information technologies (neural networks) into medicine, new opportunities arise for developing algorithms for predicting chronic rhinosinusitis recurrences [8]. The use of neural networks allows for the automation of the analysis of large volumes of clinical data and the identification of complex relationships between various factors (laboratory test data, radiological studies, patient history, etc.) contributing to chronic rhinosinusitis recurrences. Therefore, enhancing the capabilities of predicting recurrences and the course of chronic rhinosinusitis through the analysis of data arrays using neural networks can help optimize diagnostic and therapeutic strategies for each individual patient, corresponding to a personalized approach to patient management [9]. The concept of the dissertation work was based on the development of algorithms for predicting the risk of chronic rhinosinusitis recurrence using neural networks based on the analysis of retrospective clinical data to aid in patient segmentation by risk level. This aids in determining a rational surgical strategy and optimizing postoperative patient management. Research Objective: To develop an algorithm for predicting chronic rhinosinusitis recurrence based on retrospective clinical data of patients and computed tomography data of the nose and paranasal sinuses using neural networks.