Optimization and Technical Validation of the AIDE-MOI Fall Detection Algorithm in a Real-Life Setting with Older...

Area of Science:

• Gerontology
• Biomedical Engineering
• Wearable Technology

Background:

• Falls are a major cause of injury in the elderly, with prolonged lying time increasing severity.
• The AIDE-MOI motion sensor was initially developed for fall detection using a lab-based algorithm on young subjects.
• Improving fall detection accuracy is crucial for timely intervention and reducing fall-related harm.

Purpose of the Study:

• To optimize and validate an existing threshold-based fall detection algorithm for elderly individuals.
• To enhance the performance of the AIDE-MOI motion sensor's fall detection capabilities.
• To improve reaction times in fall events through more accurate detection.

Main Methods:

• A two-phase study involved 20 elderly subjects (mean age 86.25 years) at high fall risk (Morse score > 65).
• Real-time motion data was collected using the AIDE-MOI sensor over 125 days.
• An existing algorithm was optimized using phase one data and evaluated in phase two, analyzing 31 real falls.

Main Results:

• The optimized second-generation algorithm demonstrated a significant improvement in sensitivity from 27.3% to 80.0%.
• Specificity also improved from 99.9957% to 99.9978%, with a reduction in false alarms.
• The algorithm effectively distinguished between 'fall' and 'non-fall' events in elderly subjects.

Conclusions:

• The enhanced fall detection algorithm significantly improves accuracy in elderly individuals.
• The AIDE-MOI sensor, with the optimized algorithm, offers a promising solution for real-time fall monitoring.
• This advancement has the potential to reduce fall consequences by enabling faster emergency responses.