摘要
目的由于帕金森病冻结步态的突发性,临床上对其进行评估存在一定困难,为此本文研究了一种用于实时监测冻结步态的系统。方法该系统由可穿戴设备和配套的APP两部分构成,其中设备通过惯性传感器和超声波传感器采集患者腿部运动的加速度和抬脚高度数据,并传输至APP软件中,通过软件中的冻结步态识别模型进行分析。系统为构建冻结步态识别模型,首先通过实验采集12位患者的运动数据,然后经过信号预处理、特征提取和机器学习算法训练出模型,最后通过对数据集采用十折交叉验证来评估模型的准确度和精确度。结果系统对冻结步态的识别准确率可达98.6%,精确率达97.2%。结论该系统能够实时监测帕金森病患者日常生活中的冻结步态发作情况,为医生的诊疗提供定量、可靠的参考依据。
Objective Due to the unexpected nature of freezing of gait(FOG) in Parkinson disease, there are some difficulties in clinical assessment. This paper investigates a system for real-time monitoring of freezing of gait in Parkinson disease. Methods The system consists of wearable device and supporting APP. The device collects the patient’s acceleration and foot height data of the leg movement by inertial sensors and ultrasonic sensors. Then the data are transmitted to the APP software and analyzed by the FOG recognition model in the software. In order to construct the FOG recognition model, we first acquire the motion data of 12 patients through experiments and then train the model through signal preprocessing, feature extraction and machine learning algorithms. Finally, we use the ten-fold cross-validation of the data set to evaluate the accuracy and precision of the model. Results The identification accuracy of the system for freezing of gait reachs 98.6% and the precision rate is 97.2%. Conclusions The system can monitor the situation of FOG in patients’ daily life, which can provide a quantitative and reliable reference for the diagnosis and treatment.
作者
袁心一
王家莉
仇一青
杨春晖
胡小吾
吴曦
沈林勇
YUAN Xinyi;WANG Jiali;QIU Yiqing;YANG Chunhui;HU Xiaowu;WU Xi;SHEN Linyong(School of Mechatronics Engineering and Automation, Shanghai University, Shanghai 200072;Changhai Hospital, Shanghai 200433)
出处
《北京生物医学工程》
2019年第2期182-189,共8页
Beijing Biomedical Engineering
基金
国家自然科学基金(51275282)
教育部博士点基金(20123108110009)资助
关键词
帕金森病
冻结步态
可穿戴设备
特征提取
机器学习
Parkinson disease
freezing of gait
wearable device
feature extraction
machine learning