Cancer Prevention Research Centre, School of Public Health, The University of Queensland
Memorial DayOur office will be closed Monday, May 30th in observance of Memorial Day. We will reopen at regular business hours on Tuesday, May 31st.
Using Bluetooth proximity sensing to determine location in a workplace
- Published on Jun 21, 2017
Most wearable devices that measure sedentary and active time in workplaces cannot determine the context of that time. The ActiGraph wGT3X-BT and Link models allow users to not only measure movement but the wearer’s location relative to a beacon using Bluetooth proximity detection. This function has yet to be independently validated. Knowing where workers are sitting and moving at work can inform strategies suitable for workplace interventions. Twenty-five office workers (32% men, mean ± SD age 39 ± 11 years) wore a wearable camera (video recording) and the ActiGraph Link, initialised as a receiver, attached to the thigh for one work day (6.2 ± 1.1 hours). Link devices initialised as beacons were placed in the entry (n=1), kitchen (n=1), photocopy room (n=1), corridors (n=2?4), and the wearer?s office (n=2). RSSI signals from all beacons were converted to binary outcomes (1=present, 0=absent). Link-determined location was decided using two methods. Method 1: Presence/absence of signal at a single beacon location. Method 2: Signal presence was summed over a 50 s centred moving window for all beacon locations. A single location was assigned based on majority vote and time-use probabilities. Location determined by each method was checked against camera location for sensitivity, specificity and accuracy. Median sensitivity/specificity/accuracy for the office location was 99%/29%/55% (method 1) and 99%/77%/98% (method 2). The median sensitivity/specificity/accuracy for the other locations ranged from 41%/56%/55% (corridors) to 87%/65%/65% (kitchen) for method 1 and 17%/99%/99% (entry) to 83%/99%/99% (kitchen) for method 2. The ActiGraph proximity detection function shows promise as a tool for determining where workers spend time within office-based work settings. When using multiple beacons, a rolling window algorithm that chooses a single location can improve classification accuracy. This information will be for researchers planning workplace sitting interventions.
- Bronwyn Clark 1
- Suleeporn Tinakorn na ayudhaya 1
- Elisabeth Winkler 1
- Charlotte Brackenridge 1
- Genevieve Healy 1
- Stewart Trost 2
School of Exercise and Nutrition Sciences, Queensland University of Technology
ICAMPAM 2017 Abstract Booklet