Evaluation of the autonomic nervous system for fall detection

Area of Science:

• Gerontology
• Biomedical Engineering
• Physiology

Background:

• The elderly population is rapidly growing, increasing the risk of underreported health issues like falls.
• Current accelerometer-based fall detection devices have limitations in accuracy.
• Monitoring autonomic nervous system activity (ANS) may enhance fall detection capabilities.

Purpose of the Study:

• To investigate the utility of quantifying ANS activity for improving fall detection in the elderly.
• To develop and evaluate a wearable device for ambulatory ANS monitoring.
• To classify simulated falls and standing-lying transitions using physiological data.

Main Methods:

• Collected heart rate variability and electrodermal response data from 7 adult subjects during simulated falls and transitions.
• Utilized a wearable ambulatory device for continuous physiological monitoring.
• Implemented a Support Vector Machine (SVM) classification algorithm with a Polynomial kernel (p=5).

Main Results:

• The SVM model achieved a sensitivity of 70.37% and a specificity of 80% in classifying falls.
• The positive predictive value for fall detection was 73.8%.
• The classification method demonstrated good performance in distinguishing falls from normal movements.

Conclusions:

• Quantifying ANS activity via wearable devices shows promise for enhancing fall detection systems.
• SVM classification of heart rate variability and electrodermal response can effectively identify falls.
• This approach offers a potential improvement for the safety and care of the elderly population.