Hoehn-Yahr分级是现在临床上通用的对帕金森病分级的标准.基于运动传感器的可穿戴设备为帕金森病患者的运动功能评价提供了更客观和精准的监测.本文针对帕金森病的自动分级提出了一种基于六轴加速度与角速度传感器数据的自动分级算法.该算法采用基于各个动作特征的特殊运动参数和对每个运动无特异性的统计参数来共同建模.得到运动参数后,使用3个目前最先进的机器学习算法:支持向量机、K最邻近以及随机森林进行分类精度的比较.同时也分析了各个分类器使用不同参数对分类精度的影响.本研究在67例个体下的最终分类精度为89.55%.%The Hoehn-Yahr is the standard for the classification of Parkinson’s disease at present. Wearable devices based on motion sensors provide more objective and accurate monitoring for motor function evaluation of patients with Parkinson’s disease. This paper proposed an automatic grading algorithm based on six-axis acceleration and angular velocity sensor data for automatic classification of Parkinson’s disease. The algorithm used a combination of special motion parameters based on individual motion features and statistical parameters that were non-specific for each motion to model together. After obtaining the motion parameters, comparison of classification accuracy were conducted by using three current state-of-the-art machine learning algorithms such as support vector machine, K nearest neighbor, and random forests. At the same time, the influence of different parameters using different parameters on the classification accuracy was also analyzed. The final classification accuracy of the study in 67 individuals was 89.55%.
展开▼
