Electronic Thesis and Dissertation Repository


Master of Engineering Science


Electrical and Computer Engineering


Trejos, Ana Luisa


Musculoskeletal (MSK) conditions are a leading cause of pain and disability worldwide. Rehabilitation is critical for recovery from these conditions and for the prevention of long-term disability. Robot-assisted therapy has been demonstrated to provide improvements to stroke rehabilitation in terms of efficiency and patient adherence. However, there are no wearable robot-assisted solutions for patients with MSK injuries. One of the limiting factors is the lack of appropriate models that allow the use of biosignals as an interface input. Furthermore, there are no models to discern the health of MSK patients as they progress through their therapy.

This thesis describes the design, data collection, analysis, and validation of a novel muscle health model for elbow trauma patients. Surface electromyography (sEMG) data sets were collected from the injured arms of elbow trauma patients performing 10 upper-limb motions. The data were assessed and compared to sEMG data collected from the patients' contralateral healthy limbs. A statistical analysis was conducted to identify trends relating the sEMG signals to muscle health.

sEMG-based classification models for muscle health were developed. Relevant sEMG features were identified and combined into feature sets for the classification models. The classifiers were used to distinguish between two levels of health: healthy and injured (50% baseline accuracy rate). Classification models based on individual motions achieved cross-validation accuracies of 48.2--79.6%. Following feature selection and optimization of the models, cross-validation accuracies of up to 82.1% were achieved.

This work suggests that there is a potential for implementing an EMG-based model of muscle health in a rehabilitative elbow brace to assess patients recovering from MSK elbow trauma. However, more research is necessary to improve the accuracy and the specificity of the classification models.

Included in

Biomedical Commons