METHOD FOR CLASSIFICATION OF HUMAN MOTOR ACTIVITY Russian patent published in 2021 - IPC A61B5/374 

FIELD: data processing.

SUBSTANCE: invention relates to the field of digital processing and analysis of data and is intended for decoding signals of electrical brain activity associated with human movements. The method for classification of human motor activity consists in recording signals from the cerebral cortex using the EEG from sensors attached to the head. The signals filtered in the mu-rhythm range of 8 to 14 Hz are therein recorded and divided into two data arrays corresponding with the left and the right hemispheres. The data of each array is considered a path in the three-dimensional phase space. The complexity measures of the EEG signals are calculated. The complexity measures characterising the EEG signals of the right brain hemisphere are subtracted from the complexity measures characterising the EEG signals of the left hemisphere. Based on the difference in the complexity characteristics of the signals between the hemispheres, a conclusion is made about the corresponding movement. If determinism increases and recurrent time entropy decreases in the right hemisphere, movement of the left hand is determined. If entropy increases and determinism increases in the left hemisphere, movement of the right hand is determined.

EFFECT: achieved are determination of the time of start of movement and identification of two types of movements (left and right hand) based on numerical analysis of recurrent diagrams of EEG signals obtained from 6 leads from the left and right brain hemispheres, with low requirements for the computing power of the computer, reduced amount of preprocessing stages and classification results easier to interpret from a physiological point of view.

1 cl, 4 dwg