Neurocomputational identification of order parameters in gerontology

Abstract

The fallacy of using a neuroemulator only once or with a small number of iterations (p ≤ 50) to solve the group-separation problem (a binary classification problem) in a five-dimensional phase space has been demonstrated using the example of the parameters of five active components (of the 14 that were registered) of the state vector of the cardiorespiratory system in Khanty (indigenous people of Yugra, Russia) women from three age groups. The necessity of repeating the neuroemulator-based solution of the binary classification problem at least 1000 times has been demonstrated: in this case, the most significant diagnostic features, x_i, could be identified with a precision of two significant fraction digits that are most relevant for the diagnostics of the aging rate (finding a solution of the system-synthesis problem in gerontology).