An elderly fall detection using a wrist-worn accelerometer and barometer.

Area of Science:

• Gerontology
• Biomedical Engineering
• Wearable Technology

Background:

• The global population is aging, increasing the incidence of falls among the elderly.
• Falls in older adults lead to significant physical and mental health consequences.
• Existing fall detection systems often require improvement in accuracy and response time.

Purpose of the Study:

• To develop and evaluate an automatic fall detection system for the elderly.
• To integrate a 3-axis accelerometer and barometer into a wrist-worn device for fall detection.
• To address challenges with noisy sensor data and limited training datasets.

Main Methods:

• Utilized a wrist-worn device incorporating a 3-axis accelerometer and a barometer.
• Developed pre-processing and feature extraction techniques to handle noisy barometer signals.
• Explored the use of free-falling events to augment training data for the fall detection algorithm.

Main Results:

• The proposed method achieved high classification performance in detecting falls.
• The system demonstrated effectiveness across various activities and simulated falls.
• Identified a minor issue with false alarms during specific movements like sitting down.

Conclusions:

• The integrated accelerometer-barometer system is a promising approach for reliable elderly fall detection.
• The method effectively handles noisy sensor data and utilizes free-fall events for better training.
• Further refinement is needed to eliminate false alarms in specific scenarios.