SisFall: A Fall and Movement Dataset.

Area of Science:

• Biomedical Engineering
• Gerontology
• Wearable Technology

Background:

• Limited publicly available datasets hinder research in wearable fall and movement detection.
• Existing datasets often lack objective populations, diverse activities, and comprehensive data.

Purpose of the Study:

• To introduce a novel, comprehensive dataset for fall and activities of daily living (ADL) detection using a custom wearable device.
• To provide a benchmark dataset for evaluating fall detection algorithms.

Main Methods:

• Developed a wearable device with accelerometers and a gyroscope.
• Collected data from young adults and elderly individuals performing 19 ADLs and 15 fall types.
• Utilized feature extraction and threshold-based classification for initial testing.

Main Results:

• Achieved up to 96% accuracy in fall detection using the new dataset.
• Identified specific activities where most classification errors occurred.
• Observed significantly reduced fall detection performance with elderly participants, validating prior findings.

Conclusions:

• The presented dataset is a valuable resource for advancing fall and movement detection research.
• The findings highlight areas for focused development of new detection algorithms.
• The dataset's validation with elderly populations underscores the need for age-specific strategies.