A Novel Algorithm for Determining the Contextual Characteristics of Movement Behaviors by Combining Accelerometer...

Area of Science:

• Health behavior science
• Ubiquitous computing
• Sensor networks

Background:

• Global physical activity promotion efforts have faced challenges.
• Understanding the context (where, when, with whom) of behavior is key for interventions.
• Novel approaches are needed to identify effective behavior change targets.

Purpose of the Study:

• To develop and implement an innovative algorithm for determining physical activity location.
• To utilize proximity sensors and physical activity monitors for context sensing.
• To enhance the understanding of physical activity behavior in specific environments.

Main Methods:

• Deployed 19 Bluetooth beacons in a multilevel building.
• Utilized 4 wrist-worn sensors on a roving technician.
• Analyzed data using a novel algorithm with graph generation and Bayesian filters across 4 trials.

Main Results:

• The algorithm accurately predicted indoor location (R²=.9780).
• Robustness improved with longer dwelling times (>100s, error <10%).
• Device sampling rate impacted accuracy during area transitions.

Conclusions:

• The developed algorithm accurately predicts individual location within indoor environments.
• This context-sensing approach opens new research avenues for promoting healthy behaviors.
• Findings support applications in optimizing patient care and healthcare planning.