Исследование и разработка методов автоматического поиска признаков болезни Паркинсона и эссенциального тремора на базе AUC-диаграмм

Abstract

Methods and optimization algorithms for automatic search for AUC-diagram-based features of Parkinson's disease and essential tremor were studied and developed. AUC diagrams are a new method for statistical analysis of biomedical signals, based on visualizing the parameters of wave train electrical activity in the brain and muscles. The effectiveness of this method has been demonstrated in solving problems of early and differential diagnosis of Parkinson's disease and essential tremor. The disadvantage of this method is the need to construct and analyze a large number of graphic diagrams. In this regard, automation of the analysis of AUC diagrams is an urgent task. The mathematical problem of finding features based on the analysis of AUC diagrams is reduced to an optimization problem in a multidimensional feature space. A distinctive feature of the space constructed using AUC diagrams is the presence of relatively large compact areas containing local maxima and minima. This property of the feature space facilitates the search for solutions to the optimization problem, but at the same time requires the selection of optimization algorithms and fitness functions that increase the likelihood of detecting global extrema. In this work, methods for automatically searching for global extrema in the multidimensional space of features of wave train electrical activity are investigated and developed.