A fully integrated, standalone stretchable device platform with in-sensor adaptive machine learning for...

Area of Science:

• Biomedical Engineering
• Wearable Technology
• Machine Learning

Background:

• Continuous monitoring of vital and muscle activities is crucial for post-surgical throat treatments.
• Existing methods lack wireless, continuous analysis capabilities directly from throat skin.

Purpose of the Study:

• To design and validate a fully integrated, standalone stretchable device for wireless monitoring and analysis of throat activities.
• To enable machine learning-based classification of motion and speech features from diverse physiological signals.

Main Methods:

• Development of a stretchable device platform with a modified composite hydrogel for low contact impedance and adhesion.
• Integration of a triaxial broad-band accelerometer for measuring body movements and physiological vibrations.
• Utilizing a 2D-like sequential feature extractor with fully connected neurons for data processing and classification.

Main Results:

• The device achieved high-quality, long-term monitoring of muscle electrical signals.
• Accurate measurement of both large body movements and subtle physiological activities/vibrations.
• Machine learning classification of motion/speech features exceeded 90% accuracy, adapting to noisy data and new subjects.

Conclusions:

• The developed stretchable device offers wireless monitoring and machine learning analysis for throat activities.
• This technology paves the way for wearable skin-interfaced systems for remote monitoring and treatment evaluation.
• Potential applications include managing various diseases requiring continuous physiological assessment.