METHOD FOR DIAGNOSING PARKINSON'S DISEASE BASED ON VIDEO DATA ANALYSIS USING MACHINE LEARNING

摘要

FIELD: medical technology. ;SUBSTANCE: invention relates to the field of information and communication technologies (ICT) specifically designed for medical diagnostics. A method is proposed in which video data containing one or more types of motor activity of the patient is collected and obtained; the received video data is processed; for each type of motor activity, the following is performed: a file with key points of the patient’s body is obtained for each frame of the processed video data; segments are calculated based on the received key points of the patient’s body; the relative speed and acceleration are calculated based on the calculated segments and, based on the calculations performed, a data set is obtained; the data obtained at the previous stage are segmented and time and frequency features from the segmented data set are isolated; the obtained features using one or more classifiers are processed and output data are obtained, and in the learning process the best one or more classifiers for each type of motor activity are selected; the diagnosis of Parkinson’s disease is carried out on the basis of the output data obtained for each type of motor activity. ;EFFECT: invention provides an increase in the accuracy of diagnosing Parkinson’s disease based on the analysis of video data containing the patient’s motor activity. ;16 cl, 3 dwg