Multimodal sensor-based fall detection within the domestic environment of elderly people

Area of Science:

• Gerontology
• Biomedical Engineering
• Human-Computer Interaction

Background:

• Falls pose a significant health risk to the elderly population.
• Increasing elderly populations worldwide necessitate efficient fall detection solutions.
• Current fall detection systems lack consistent reliability in real-world settings.

Purpose of the Study:

• To evaluate sensor-based fall detection accuracy using accelerometers, video, and microphones.
• To analyze fall incidents using standardized protocols and geriatric assessments.
• To assess user acceptance of fall detection sensor technology via questionnaires.

Main Methods:

• Recruited 28 community-dwelling German seniors (mean age 74.3 years).
• Utilized accelerometers, video cameras, and microphones for sensor-based fall detection.
• Collected data over 8 weeks, including 1225.7 measurement days.

Main Results:

• Fall detection algorithms identified an average of 2.66 falls per day.
• 15 falls occurred during the study period.
• The system correctly recognized 12 out of 15 falls (80% accuracy).

Conclusions:

• Sensor-based fall detection performs well in controlled environments but faces challenges in real-world application.
• While generally useful, user acceptance of fall detection sensors declined after the study.
• Further research is needed to improve the reliability and user acceptance of fall detection technologies for the elderly.