Wearable multisource quantitative gait analysis of Parkinson's diseases
Junxiao Xie,Huan Zhao,Junyi Cao,Qiumin Qu,Hongmei Cao,Wei-Hsin Liao,Yaguo Lei,Linchuan Guo
DOI: https://doi.org/10.1016/j.compbiomed.2023.107270
IF: 7.7
2023-07-21
Computers in Biology and Medicine
Abstract:As the motor symptoms of Parkinson's disease (PD) are complex and influenced by many factors, it is challenging to quantify gait abnormalities adequately using a single type of signal. Therefore, a wearable multisource gait monitoring system is developed to perform a quantitative analysis of gait abnormalities for improving the effectiveness of the clinical diagnosis. To detect multisource gait data for an accurate evaluation of gait abnormalities, force sensitive sensors, piezoelectric sensors, and inertial measurement units are integrated into the devised device. The modulation circuits and wireless framework are designed to simultaneously collect plantar pressure, dynamic deformation, and postural angle of the foot and then wirelessly transmit these collected data. With the designed system, multisource gait data from PD patients and healthy controls are collected. Multisource features for quantifying gait abnormalities are extracted and evaluated by a significance test of difference and correlation analysis. The results show that the features extracted from every single type of data are able to quantify the health status of the subjects ( p < 0.001, ρ > 0.50). More importantly, the validity of multisource gait data is verified. The results demonstrate that the gait feature fusing multisource data achieves a maximum correlation coefficient of 0.831, a maximum Area Under Curve of 0.9206, and a maximum feature-based classification accuracy of 88.3%. The system proposed in this study can be applied to the gait analysis and objective evaluation of PD.
engineering, biomedical,computer science, interdisciplinary applications,mathematical & computational biology,biology