JMIR Publications

ABSTRACT

Background:

The COVID-19 pandemic saw the implementation of public health policies aimed at restricting human mobility to curb the spread of infection. Mobility is an often-overlooked determinant of human health and is linked to both infectious and chronic diseases. The development of tools to collect mobility data can unravel the complexities of human behavior and inform public health policies. Google has capitalized on its GPS-based location tracking to capture human movement out of the house during the pandemic: Google Mobility Reports has become the gold standard in mobility research. Yet, human mobility inside the home remains relatively unexplored. Here we investigated in-house mobility data from ecobee’s smart thermostats during the COVID-19 pandemic in Canada.

Objective:

This study aimed to assess the suitability of smart thermostat data through direct comparison with Google’s residential mobility data.

Methods:

Motion sensor data was collected from the ecobee “Donate your Data” program through Google’s BigQuery cloud platform. Residential mobility data were obtained from the Google Mobility Report. The analysis focused on the Canadian provinces of Ontario, Quebec, Alberta, and British Columbia from February 15, 2020, to February 14, 2021. The data cleaning, analysis, and visualization used the MS Azure platform with Python and R coding languages. Changes in mobility above baseline were determined and compared between the two datasets. The statistical significance of the association was determined using Pearson’s correlation and Spearman’s correlation [34]. We analyzed day-by-day, week-by-week, and month-by-month seasonality patterns across both datasets and performed anomaly detection.

Results:

Results showed significant changes in week-by-week and month-by-month population mobility for the selected provinces which tracked with pandemic-related policy changes. The ecobee data was significantly associated with Google data. Analysis of Google’s day-by-day patterns found greater mobility changes on weekdays; this trend was not captured in the ecobee data. Anomaly detection found significant mobility deviations corresponding to policy changes and cultural festivities.

Conclusions:

The findings from this study demonstrate that the Canadian stay-at-home order and work-from-home policies had a significant impact on population mobility. This could be captured using both out-of-house residential mobility data from Google and in-house smart thermostat data from ecobee. Therefore, smart thermostats are a valid tool to support intelligent monitoring of population mobility in response to policy-related changes.