A large harmonized upper and lower limb accelerometry dataset: A resource for rehabilitation scientists

Area of Science:

• Rehabilitation research
• Movement science
• Wearable sensor technology

Background:

• Wearable sensors offer valuable insights into daily life movement for rehabilitation.
• Collecting and processing sensor data presents significant burdens, limiting accessibility for researchers.
• Existing datasets are often fragmented, hindering large-scale analysis and reproducibility.

Purpose of the Study:

• To present a harmonized, comprehensive wearable sensor dataset for advancing rehabilitation research.
• To address challenges in data accessibility, processing, and reproducibility in wearable sensor studies.
• To provide researchers with a valuable resource for investigating human movement and clinical conditions.

Main Methods:

• Combined data from eight studies, encompassing 2,885 recording days of upper and lower limb sensor data.
• Included data from 790 individuals across a wide age range (0-90 years) and diverse demographic backgrounds.
• Incorporated data from various clinical populations, including neurotypical individuals, stroke survivors, and those with Parkinson's disease or orthopaedic conditions.

Main Results:

• The dataset features a large and diverse participant pool with balanced sex representation.
• It includes comprehensive demographic and clinical condition information, enhancing its utility.
• The dataset is publicly available, promoting open science principles.

Conclusions:

• This harmonized dataset significantly facilitates the use of wearable sensor data in rehabilitation research and practice.
• It enhances the reproducibility and replicability of studies utilizing wearable sensor technology.
• The open-source dataset and accompanying tools minimize scientific effort and costs, accelerating research progress.