A dataset for the development and optimization of fall detection algorithms based on wearable sensors

Area of Science:

• Biomedical Engineering
• Human-Computer Interaction
• Wearable Technology

Background:

• Fall detection systems are crucial for elderly and at-risk individuals.
• Existing systems require robust datasets for accurate fall event identification.
• Wearable sensors offer a promising approach for continuous monitoring.

Purpose of the Study:

• To create and describe a comprehensive dataset of simulated falls and Activities of Daily Living (ADL).
• To facilitate the development and testing of automatic fall detection and classification algorithms.
• To advance research in wearable sensor-based health monitoring.

Main Methods:

• Acquired data from 8 subjects simulating 13 distinct fall types and 5 ADL types, each repeated thrice.
• Utilized a waist-worn Micro-Electro-Mechanical Systems (MEMS) Inertial Measurement Unit (IMU) sensor (MARG sensor).
• Recorded time-series data of acceleration and orientation (yaw, pitch, roll angles).

Main Results:

• A dataset comprising various simulated fall scenarios (forward, backward, lateral, syncope) and common ADLs was successfully generated.
• The dataset captures detailed sensor data (acceleration, orientation) during these activities.
• The data provides a valuable resource for algorithm training and validation.

Conclusions:

• The described dataset is suitable for developing and testing advanced fall detection algorithms.
• This resource can significantly contribute to improving the reliability of automated fall detection systems.
• Such systems have the potential to enhance safety and provide timely assistance to individuals prone to falls.