Increasing detection efficiency of border mental disorders based on adaptive decomposition and cepstral analysis of speech signals.

Abstract

The detection accuracy of borderline mental disorders depends on correct processing of speech signals. The main reason of low accuracy and large errors in measurements is associated with the use of inefficient and non-adaptive methods for processing of non-stationary speech signals. In this paper, the authors propose a method for increasing the detection efficiency of borderline mental disorders based on adaptive decomposition technology for non-stationary signals, namely, improved complete ensemble empirical mode decomposition with adaptive noise and melfrequency cepstral analysis. A block diagram for the method and a brief mathematical description are presented. The research results are presented, on the basis of which it was concluded that the method proposed by the authors can successfully be tested in remote monitoring systems of psychogenic disorders to accelerate the treatment process.