A Review of Statistical Analyses on Physical Activity Data Collected from Accelerometers

Area of Science:

• Biomedical Engineering
• Epidemiology
• Data Science

Background:

• Wearable accelerometer devices offer real-time, second-by-second physical activity monitoring.
• These devices generate summary metrics like energy expenditure, sedentary time, and moderate-to-vigorous physical activity.
• Advancements in technology facilitate robust research into physical activity and disease risk.

Purpose of the Study:

• To review methods for analyzing raw accelerometer data.
• To highlight R packages for generating activity-related variables.
• To discuss advanced analytical approaches for research purposes.

Main Methods:

• Review of existing literature on accelerometer data analysis.
• Identification and categorization of relevant R packages.
• Discussion of longitudinal and functional data analysis techniques.

Main Results:

• Several methods exist for processing raw accelerometer data into meaningful metrics.
• Various R packages are available to assist researchers in data analysis.
• Longitudinal and functional data approaches offer powerful tools for complex research questions.

Conclusions:

• Accelerometer data analysis is crucial for understanding physical activity's role in disease.
• The availability of specialized R packages simplifies data processing and analysis.
• Advanced statistical methods enable deeper insights into physical activity patterns and health outcomes.