Tymkiv P. Identification of parameters of the mathematical model of retinal response to low-intensity stimulation

The dissertation is devoted to the study of methods of processing low – intensity electroretinosignal, and the development of a method of preliminary data preparation for the application of the adaptive-recursive filter. It has been established that to identify the risks of toxicity of the organism (type of toxins, dose, duration of exposure, etc.) and early diagnosis, use the assessment of small characteristic changes in the electroretinosignal (ERS) or its parts (waves). To increase the accuracy and resolution, which is necessary to determine the risks of neurotoxicity, reduce the energy of light stimulation. A low-intensity ERS is obtained, which is significantly distorted by noise due to the decrease in the ratio of signal energy to noise energy. The application of known algorithms for determining the parameters of the mathematical model of ERS requires a significant calculation time, which makes it impossible to quickly reconfigure the filter when processing low-intensity ERS. To do this, we use the Hooke-Jeeves method, which is a combination of exploratory search with cyclic change of variables and sample search. When applying the algorithm based on the Hooke-Jeeves method, the time for determining the coefficients is reduced by 120 times, other things being equal. By the method of statistical testing, the reliability of the method of determining the coefficients was determined and ROC-analysis was applied, which showed higher sensitivity and specificity of the proposed method in comparison with the method of direct directed search.