Loading...

Design and Development of a Mobility Recognition System in PD Patients for Tele-rehabilitation

Mohammadi Nasrabadi, Amin | 2019

639 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 52136 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Behzadipour, Saeed; Alibiglou, Laila
  7. Abstract:
  8. Parkinson's disease is a neurodegenerative disorder that affects motor functions. Performing mobility exercises help patients slowing down the progression of the illness and improving symptoms of the disease. Assessment and evaluation of activities of mobility exercises are critical for any treatment program particularly in tele-rehabilitation system. The purpose of the current study is to design an affordable and accurate wearable device with inertial measurement units (IMUs) for mobility activity recognition in Parkinson’s patients. The optimum number and arrangement (i.e. configuration) were found to minimize the cost while maintaining a fair accuracy. The activity recognition was performed for in-home mobility exercises and the LSVT-BIG [2] exercises, which included 34 movements. The collected data was processed in windows of 2.5 seconds. Eight features in time and frequency domain, and discrete wavelet transforms were calculated. Dimension reduction was performed using PCA algorithm. NM , RBF , SVM , and k-NN classifiers were then trained and used to recognize the activity. A genetic algorithm was utilized to decide which sensors and signals take part in the classification to perform the best accuracy. The results showed that the sensors installed on the left shank, right thigh, left forearm and right arm provided a precision of 94.3% and sensitivity of 93.4% using NM classification. Also, an adaptation algorithm was utilized in order to maintain the quality of recognition for new users. In order to achieve an affordable real-time system, a multiple window classifier and a filter-approach recognition have been utiliezed to avoid signal transition. Therfore the system can recognize the activity of subjects in a real-time situation. A clothing covering the sensors and wires and a GUI have been developed for the ease of use. After all, an instruction manual for sensor installation was provided by sensitivity analysis of sensor placement
  9. Keywords:
  10. Activity Recognition ; Wearable Sensor ; Tele-rehabilitation ; Parkinson Disease ; Inertial Measurement Unite (IMU) ; Real Time System ; Mobility Model Recognition

 Digital Object List

 Bookmark

No TOC