Optimal Location for Fall Detection Edge Inferencing

Area of Science:

• Gerontology
• Biomedical Engineering
• Wearable Technology

Background:

• Falls are a leading cause of injury in the elderly, with 32-42% of those over 70 falling annually.
• Delayed medical attention after a fall can worsen injuries like concussions, traumatic brain injuries, or fractures.
• Current wearable fall detection sensors (FDS) are often worn on the neck or wrist.

Purpose of the Study:

• To determine the optimal body placement for mobile, wireless, low-power fall detection sensors (FDS) to improve accuracy in the elderly population.
• To evaluate the effectiveness of different sensor placements using machine learning models.

Main Methods:

• Collected data from Inertial Measurement Unit (IMU) sensors placed at sixteen different body locations.
• Utilized four distinct machine learning models trained on features extracted from the IMU sensor data.
• Analyzed sensor data to identify patterns associated with falls across various body placements.

Main Results:

• The study identified specific body locations yielding higher accuracy for fall detection.
• Optimal placement for a fall detection sensor (FDS) was determined to be in front of the shinbone.
• Machine learning models demonstrated varying performance based on sensor location.

Conclusions:

• Sensor placement significantly impacts the accuracy of wearable fall detection systems.
• Positioning fall detection sensors (FDS) on the shinbone offers superior detection capabilities compared to traditional placements.
• This finding can lead to more effective fall monitoring devices for elderly individuals, potentially reducing fall-related injuries.