Carroll score prediction method for hand motion function of stroke patients

Abstract

The invention discloses a Carroll score prediction method for hand motion function of stroke patients, and aims at solving the problems of disadvantages of a traditional scale evaluation mode and an automatic evaluation method based on machine learning. The prediction method comprises the following steps: acquiring data, preprocessing the data, performing space-time feature extraction on hand motion data by using a convolutional neural network with grouping constraint for, and automatically generating a Carroll score. The method realizes automatic score prediction of hand motion function of stroke patients based on the Carroll score scale. A designed hand motion data acquisition system realizes the objective measurement of the hand motion function, eliminates the subjective deviation caused by a traditional manual observation scoring mode; and the method carries out the space-time feature extraction and the Carroll score prediction on the multi-channel sensor sequence data by using a depth learning way, reduces the intervention of human factors in the manual feature extraction process, and improves the automatic evaluation precision.

Description

technical field [0001] The invention relates to the application field of deep learning, in particular to a method for predicting Carroll score of hand motor function after stroke. Background technique [0002] The hand is an important organ for the human body to communicate with the outside world, and it is also the most complex and delicate organ of the human body. The main function of the hand is grasping, which is divided into force grasping (spherical grasping, columnar grasping and pulling) and fine grasping (fingertip pinch, finger pulp pinch, side pinch and three-finger pinch). Stroke patients often have upper limb hand dysfunction due to nervous system damage, and the speed and accuracy of hand operation are reduced. Grip strength and coordination ability are also directly related to the recovery of motor function of the individual patient. Therefore, the evaluation of hand motor function can provide a basis for the formulation of rehabilitation programs for patient...

AI Technical Summary

Problems solved by technology

[0005] The traditional scale assessment method requires physicians to participate in the whole process, and there are the following problems: 1) The hand assessment scale contains a large number of assessment items, and the assessment process is time-consuming and inefficient; 2) The assessment process requires real-time supervision and guidance of professional assessors; 3 ) The evaluation results are greatly influenced by the subjective evaluation of the evaluators, and there are subjective differences; 4) There are differences in the selection of evaluation scales in different regions and institutions, which is not conducive to clinical and scientific exchanges

[0006] The automatic evaluation method based on machine learning manually extracts the time domain and frequency domain features of the collected data in the same way, ignoring the differences of the signals themselves and the dependencies between the signals, and the insufficient feature extraction due to subjective experience, which affects the classification algorithm performance, making the evaluation results less accurate

Images