A Wearable System for Monitoring Neurological Disorder Events with Multi-Class Classification Model in Daily Life

Area of Science:

• Biomedical Engineering
• Neurology
• Signal Processing

Background:

• Dysphagia and dysarthria are common neurological disorder symptoms.
• Current monitoring methods are cumbersome and clinic-bound.
• Continuous symptom tracking is vital for rehabilitation.

Purpose of the Study:

• To develop an innovative wearable system for detecting neurological disorder events.
• To utilize a mechano-acoustic (MA) sensor for sensitive, noise-resilient signal capture.
• To implement a multi-class ensemble classification model for accurate symptom discernment.

Main Methods:

• A novel wearable mechano-acoustic (MA) sensor was designed to detect neck vibrations.
• A multi-class ensemble classification model, an ensemble neural network, was developed.
• The model was trained on both raw waveforms and mel spectrograms of recorded signals.

Main Results:

• The proposed system achieved a high classification accuracy of 91.94%.
• The MA sensor demonstrated high sensitivity to relevant vibrations, unaffected by ambient noise.
• The ensemble model outperformed previous single neural network approaches.

Conclusions:

• The developed wearable system offers a promising, accurate, and portable solution for monitoring dysphagia and dysarthria.
• This technology can facilitate continuous, real-world symptom assessment for neurological disorder patients.
• The findings support the integration of wearable sensors and AI in neurological rehabilitation.