Skin-Interfaced Wireless Biomechanical System for Detecting Aspiration in Elderly Populations

Area of Science:

• Biomedical Engineering
• Gerontology
• Machine Learning

Background:

• Aspiration is common in the elderly, leading to serious health issues like pneumonia.
• Current diagnostic methods (VFSS, FEES) are invasive, complex, or use radiation.
• There's a need for continuous, non-invasive aspiration monitoring.

Purpose of the Study:

• To develop a skin-interfaced wireless biomechanical system for continuous, non-invasive aspiration detection.
• To utilize machine learning algorithms for automated aspiration identification.
• To validate the system's efficacy in a nursing-home setting.

Main Methods:

• Developed a wireless, skin-interfaced biomechanical sensor system.
• Implemented machine learning algorithms to detect post-swallow choking coughs.
• Evaluated the system on a cohort of elderly individuals (14 aspiration, 64 non-aspiration).

Main Results:

• Achieved 92% average accuracy across four activity types using leave-one-out cross-validation.
• Demonstrated 91% sensitivity in recognizing aspiration events.
• Showed patient-level generalization in additional testing on unseen aspiration patients.

Conclusions:

• The developed system provides a user-friendly, non-invasive solution for continuous aspiration monitoring.
• The technology holds promise for improving elderly healthcare in both home and institutional settings.
• Enhanced interpretability through reliability analysis supports clinical deployment.