An Unobtrusive Fall Detection System Using Ceiling-mounted Ultra-wideband Radar

Area of Science:

• Gerontology
• Biomedical Engineering
• Signal Processing

Background:

• Falls pose significant risks to older adults' health and independence.
• Existing fall detection systems often lack unobtrusiveness or are prone to occlusion.
• There is a need for reliable, automatic fall detection to ensure timely assistance.

Purpose of the Study:

• To develop and evaluate an unobtrusive fall detection system utilizing ultra-wideband (UWB) radar.
• To enhance motion detection and reduce noise in UWB data through innovative pre-processing.
• To assess the system's performance in detecting simulated falls and activities of daily living (ADL) in older adults.

Main Methods:

• A ceiling-mounted UWB radar system was implemented to minimize object occlusion.
• An innovative pre-processing technique was applied to raw UWB data for motion enhancement and noise reduction.
• A convolutional neural network (CNN) algorithm was trained and tested using data from ten participants performing simulated falls and ADLs in varied environments.

Main Results:

• The UWB radar fall detection system achieved a high overall accuracy of 93.97%.
• The system demonstrated excellent performance with a sensitivity of 95.58% and specificity of 92.68%.
• The developed pre-processing method effectively enhanced relevant motion signals and reduced environmental noise.

Conclusions:

• The proposed UWB radar system offers a promising unobtrusive solution for automatic fall detection in older adults.
• The CNN-based algorithm, coupled with advanced pre-processing, provides accurate and reliable fall identification.
• This technology has the potential to significantly improve safety and reduce the impact of falls on the elderly population.