Classifying and tracking rehabilitation interventions through machine-learning algorithms in individuals with stroke.

Area of Science:

• Biomedical Engineering
• Rehabilitation Science
• Machine Learning

Background:

• Stroke is a primary cause of long-term disability globally.
• Rehabilitation therapy is crucial for post-stroke recovery.
• Tracking in-home rehabilitation is challenging due to resource limitations.

Purpose of the Study:

• To develop and evaluate a methodology for classifying and tracking rehabilitation interventions for stroke survivors.
• To address the gap in monitoring in-home rehabilitation in resource-limited environments.

Main Methods:

• Developed personalized classification algorithms, including neural networks.
• Utilized accelerometry-based wearable sensors on limbs to collect movement data.
• Applied algorithms to classify four distinct rehabilitation exercises.

Main Results:

• Classification accuracy ranged from 64% (traditional algorithms) to 94% (neural network algorithms).
• A novel method was introduced to assess bilateral mobility changes over the therapy duration.

Conclusions:

• Personalized supervised learning algorithms can effectively classify and track rehabilitation activities.
• This approach is viable for monitoring functional outcomes in low- and middle-income countries (LMICs).