A Large Harmonized Upper and Lower Limb Accelerometry Dataset: A Resource for Rehabilitation Scientists

Area of Science:

• Rehabilitation research
• Biomedical engineering
• Wearable technology

Background:

• Wearable sensors are crucial for measuring daily movement in rehabilitation research.
• Collecting and processing sensor data presents significant burdens, limiting accessibility for scientists.
• Existing datasets are often fragmented and lack standardization.

Purpose of the Study:

• To present a harmonized, comprehensive wearable sensor dataset for rehabilitation.
• To facilitate broader access to sensor data for research and clinical practice.
• To improve reproducibility and reduce costs in wearable sensor studies.

Main Methods:

• Combined 2,885 recording days of upper and lower limb sensor data from eight studies.
• Included 790 individuals (ages 0-90) with diverse demographics and clinical conditions.
• Dataset is publicly available with open-source code and an interactive app.

Main Results:

• A large-scale, diverse dataset of wearable sensor data for movement analysis.
• Publicly accessible data and tools to lower barriers for researchers.
• Facilitation of advanced rehabilitation research and practice.

Conclusions:

• The harmonized dataset and accompanying tools will accelerate progress in rehabilitation.
• Enhanced accessibility and standardization will improve the quality of wearable sensor research.
• This resource aims to minimize duplicated efforts and costs in the field.