Using functional principal component analysis (FPCA) to quantify sitting patterns derived from wearable sensors.

Area of Science:

• Biomedical Engineering
• Public Health
• Data Science

Background:

• Sedentary behavior (SB) is a significant risk factor for chronic diseases.
• ActiGraph and activPAL are common wearable accelerometers used in SB research.
• Understanding movement variation during sitting is crucial for health assessment.

Purpose of the Study:

• To quantify movement patterns and variations during sitting events identified by activPAL.
• To examine the association between these movement patterns and health outcomes like blood pressure.
• To apply Multilevel Functional Principal Component Analysis (MFPCA) for analyzing sedentary behavior.

Main Methods:

• 314 overweight postmenopausal women wore ActiGraph and activPAL for a week.
• Minute-level time-series data were processed from both devices.
• MFPCA was used to analyze ActiGraph activity counts during activPAL-identified sitting bouts.

Main Results:

• Two subject-level and six day-level principal components explained over 90% of activity count variation.
• Movement patterns during sitting were characterized by fluctuation and duration of sitting bouts.
• Higher diastolic blood pressure was associated with lower activity counts during sitting.

Conclusions:

• MFPCA effectively identifies variations in movement patterns during sitting.
• These sedentary behavior patterns are associated with cardiovascular health.
• MFPCA offers a novel statistical approach for analyzing SB patterns and their health implications.