Longitudinal Associations Between Timing of Physical Activity Accumulation and Health: Application of Functional Data...

Area of Science:

• Biomedical Engineering
• Public Health
• Data Science

Background:

• Accelerometers provide high-resolution physical activity (PA) data, crucial for understanding human movement.
• Traditional analysis often relies on day-level summaries, potentially missing nuanced temporal patterns.

Purpose of the Study:

• To apply functional principal component analysis (FPCA) for analyzing temporal PA patterns in overweight/obese women.
• To investigate the association between these PA patterns and obesity-related health outcomes over one year.

Main Methods:

• Longitudinal functional principal component analysis (FPCA) was used to decompose high-frequency PA data.
• Mixed-effects regression models were employed to link PA patterns with health outcomes.
• Data from 245 overweight/obese women across three visits over a 1-year period were analyzed.

Main Results:

• Significant associations were found between PA variation (subject and visit-level) and health outcomes.
• The timing of daily physical activity was identified as a key factor influencing changes in health outcomes.
• These detailed temporal insights were not achievable with conventional day-level PA summaries.

Conclusions:

• Longitudinal FPCA effectively elucidates temporal patterns in high-frequency PA data.
• Understanding PA patterns, including daily timing, is vital for developing effective weight-loss guidelines.
• This approach offers a more comprehensive view of PA-health outcome relationships.