الفهرس 
IntroductionOnly 14 pages are availabe for public view
 |  | from | 144 |  |  |
| | | from | 144 | | |
Abstract
Stroke is a world wide spread disease being the third cause of movement disability for adults and the second cause of death in the world. The effects of stroke are long lasting, severe and disabling and significantly affect the patient’s quality of life as it restricts their ability to carry on their daily life independently It is estimated that each year around 15 million people suffer from a stroke and approximately 1 of each 6 persons are likely to be diagnosed with stroke every year. This not only has a tremendous impact on the lives of patients and their loved ones but also represents an enormous burden for the healthcare system around the world as it causes a huge cost to society due to increased treatment numbers of post stroke patients.
With the need of an uninterrupted rehabilitation journey for the patients as studies showed that the highest recovery rates are achieved within the first six months, with the movement disabilities after this period high tendency to be a chronic movement function disability, and also with restrictions like rehabilitation cost increase due to the increased demand on the health care system, time and place restrictions of patients living in rural areas that makes the rehabilitation sessions for them hard to achieve due to the lack of health care centers near them, or even restrictions enforced by uncontrolled natural events like the pandemics and the spread of COVID-19 which put people’s life in danger exposing to other humans, thus enforcing the isolation of patients and now allowing them to continue their rehabilitation session, thus holding the rehabilitation journey and increasing the chance of their impairment to become a chronic one; the need for telerehabilitation arises, removing constraints and allowing the patient to have a low-cost, effective rehabilitation at the time and place of their choice.
Sensor based telerehabilitation systems showed the potential and reliability in automat- ing the assessment of the patient’s performance in performing the exercises, thus we propose in this study a sensor based automated assessment system that uses various machine learning models to assess the quality of the movement done by the patient and finally display the score of the movement to the patient in order for the patient to be able to keep progress or share the results with the physiotherapists for supervision and intervention if needed. The system is based on low-cost sensors: Microsoft Kinect sensor, and a MYO armband; the movement data is collected using those 2 sensors and features are extracted to be tested using machine learning models, using One-class SVM the system was able to achieve 86.6% accuracy in successfully classifying the movement according to FMA assessment method, while the Random Decision Forest achieved a higher accuracy of 94% showing the reliability of the designed system.