Data Sharing

Databrary - Sharing Videos of Falls in Long Term Care

In 2018, the IPML began sharing a unique collection of videos of real-life falls experienced by older adults in two long term care facilities in the Vancouver area. The videos are shared for research and education via Databrary, an NYU-based data sharing network.

Please contact Steve Robinovitch (at stever@sfu.ca) if you are interested in gaining access to the video dataset. 

DATA DESCRIPTION:

The dataset contains 300 videos of falls experienced by 118 adults (57 women who accounted for 175 falls, and 61 men who accounted for 125 falls). At the time of falling, the individuals ranged in age from 58 years to 98 years (mean age = 82.8 years, SD = 7.6 years). All individuals captured falling were residing in two partnering long-term care facilities in British Columbia, where the falls occurred. All falls occurred in common areas of the long-term care homes (lounges, dining rooms and hallways). No cameras were installed in bedrooms or bathrooms.

STRUCTURE OF THE DATASET:

Each folder contains a video of the fall, and an Excel file (.xlsx) with information on the demographic and clinical characteristics of the indivdiual who experienced the fall, along with details on the number of camera views, frame rate and resolution of the video. 

 

Shared Inertial Measurement Unit Dataset

SFU TIPS researchers are now publicly sharing a comprehensive set of kinematic data from body-worn sensors (accelerometers, gyros and magnetometers) acquired during laboratory experiments of falls, near-falls, and activities of daily living, to use in the development and testing of automatic fall detection sensor systems. The dataset can be downloaded here.

Inertial Measurement Unit Fall Detection Dataset (IMU Dataset) is a dataset devised to benchmark fall detection and prediction algorithms based on acceleration, angular velocity and magnetic fields of body-worn APDM Opal IMU sensors recording at 128 Hz at 7 body locations (right ankle, left ankle, right thigh, left thigh, head, sternum, and waist). Detailed description of the dataset and column names are in README.txt file.

Use of this dataset in publications must be acknowledged by referencing the following publication:
- Omar Aziz, Magnus Musngi, Edward J. Park, Greg Mori, Stephen N. Robinovitch. "A comparison of accuracy of fall detection algorithms (threshold-based vs. machine learning) using waist-mounted tri-axial accelerometer signals from a comprehensive set of falls and non-fall trials". Med Biol Eng Comput (2017) 55: 45.
We also appreciate if you drop us an email (at stever@sfu.ca) to inform us of any publication using this dataset.