Physical activity phenotyping with activity bigrams, and their association with BMI

Area of Science:

• Biomedical research
• Data science
• Public health

Background:

• Traditional physical activity analysis relies on simple metrics like average activity counts and time spent in different intensity zones.
• These standard measures often fail to capture the dynamic nature of physical activity patterns.
• This study introduces a novel approach using 'bigrams' from text mining to analyze activity transitions.

Purpose of the Study:

• To introduce and validate the use of activity bigrams for characterizing physical activity.
• To explore the association between these novel activity bigram variables and body mass index (BMI).

Main Methods:

• Utilized data from 4810 participants in the Avon Longitudinal Study of Parents and Children (ALSPAC).
• Generated bigram frequency profiles for each participant, representing transitions between activity states.
• Tested associations between bigram frequencies and BMI, controlling for overall activity levels.

Main Results:

• Found significant associations between specific activity bigram frequencies and BMI.
• A decrease in sedentary-to-moderate or moderate-to-sedentary transitions, coupled with an increase in moderate-to-vigorous or vigorous-to-moderate transitions, was linked to a 2.36 kg/m2 lower BMI.
• These findings were adjusted for time spent in sedentary, low, moderate, and vigorous activity.

Conclusions:

• Activity bigrams offer a new way to quantify how physical activity levels change sequentially.
• These novel variables can reveal associations between dynamic activity patterns and other health-related traits.
• This approach enhances the understanding of physical activity beyond simple summary statistics.