Milyan N. Models and methods of decision making in the creation of clinical diagnostic systems.

The thesis is devoted to the solving of a relevant scientific problem of the development of methods of classical, deep, and scientific machine learning under uncertainties with the use of the minimax approach. A cyber-physical model for biosensor systems is considered, which is based on a mathematical description of the continuous dynamics of populations in combination with the dynamic logic used for discrete events. The study of stability is based on the algorithm of scientific machine learning, which offers a solution to classifying cyber biosensor systems. The algorithm of scientific machine learning includes the following stages: the definition of system trajectories classes; the generation of a matrix of random initial values of time delay and speed parameters; the model implementation and classification of input data; the construction of a matrix of connections between initial values, the time delay and between speed parameters; the application of the induction tree of the C5.0 algorithm to the relationship between the initial values and velocity parameters.