BAYESIAN HIERARCHICAL MODELING AND ANALYSIS FOR ACTIGRAPH DATA FROM WEARABLE DEVICES.

Area of Science:

• Public Health
• Biostatistics
• Epidemiology

Background:

• Low physical activity levels in Americans contribute to preventable diseases like diabetes, hypertension, and heart conditions.
• Monitoring human activity is crucial for developing interventions linked to environmental factors that promote physical activity.
• Wearable devices (actigraph units) generate high-resolution data on gross motor activity, necessitating advanced analytical approaches.

Purpose of the Study:

• To develop a Bayesian hierarchical model for analyzing spatial-temporal actigraphy data.
• To estimate physical activity levels along specific trajectories.
• To identify trajectories and spatial zones associated with higher physical activity and predict activity in new trajectories based on health attributes.

Main Methods:

• Utilized a Bayesian hierarchical modeling framework to analyze spatial-temporal actigraphy data.
• Incorporated subject-level health attributes and spatial-temporal dependencies into the model.
• Applied the framework to data from the Physical Activity through Sustainable Transport Approaches in Los Angeles (PASTA-LA) study.

Main Results:

• Developed a model for fully model-based inference on activity trajectories.
• Identified specific spatial zones and trajectories associated with significantly higher physical activity levels.
• Accounted for heterogeneity in physical activity patterns across different subjects and environments.

Conclusions:

• The proposed Bayesian framework effectively analyzes spatial-temporal actigraphy data to understand physical activity patterns.
• This approach can inform targeted public health interventions by identifying environments that encourage physical activity.
• The findings highlight the importance of considering spatial and temporal factors, along with individual health attributes, in physical activity research.