Research from the University of Waterloo has made it possible to accurately track indoor temperatures to determine households with life-threatening conditions.
While there is legislation in Canada on heating public buildings and rental units as it gets colder outside, no rules exist for cooling during extreme heat.
Researchers from the Ubiquitous Health Technology Lab (UbiLab) at Waterloo tested smart thermostats with Wi-Fi to automatically control heating and cooling settings to collect data on indoor temperatures across North America from 2015 to 2024. The team then used machine-learning algorithms to accurately track indoor temperatures to determine households with life-threatening conditions.
"During the extreme B.C. heat storm of the summer of 2021, almost all of the 619 deaths - 98 per cent of them - occurred inside the home," said Dr. Jasleen Kaur, an adjunct professor at Waterloo. "We are working to predict vulnerability in terms of which homes are most at risk by identifying the hot spots or regions that are most likely to require emergency services."
Heat-related health risks can be severe and even deadly for vulnerable populations who spend much of their time indoors, including older adults, people with pre-existing conditions, people with disabilities and individuals who are pregnant.
The UbiLab researchers are developing real-time dashboards that would be available online to the public, in addition to an early-warning system. This information could identify neighbourhood hotspots for agencies in order to inform government decision-making and emergency measures.
A prototype of the dashboard researchers are developing as part of a heat warning system. (University of Waterloo)
Dr. Plinio Morita, a professor in Public Health Sciences and director of UbiLab, said early detection and accurate forecasting of high indoor temperatures during these events are critical in preventing heat-related injury and death.
Experts define temperatures of 26 degrees for a continuous eight hours or more as a level one threat. At 31 degrees for more than eight hours, it is a level two threat, where residents are at the highest health risk.
Recognizing that some households may not be able to afford smart thermostats, Morita said there is significant potential for public health programs and government subsidies to bridge this gap.
"By predicting dangerous indoor conditions, local health systems and authorities can prioritize their resources and interventions more efficiently, ensuring that those most at risk receive timely support," Morita said. "Using AI to make these predictions helps ensure resources are directed in places such as low-income communities and long-term care facilities that don't have air conditioning."
Once an alert identifies vulnerable households, authorities could know where to implement both immediate and long-term measures. These interventions include establishing cooling strategies by distributing fans, for example, in addition to deploying emergency health services.
The team says in the long-term, subsidies for smart thermostats and investments in green infrastructure can help reduce heat risks.
"The widespread adoption of smart thermostats in Canadian homes and the prevalence of social media provide valuable opportunities to address existing data and knowledge gaps," Morita said. "However, the persistence of climate misinformation and denial challenges public health communication efforts."
The study, Deep Learning Models for Health-Driven Forecasting of Indoor Temperatures in Heat Waves in Canada: An Exploratory Study Using Smart Thermostats, appears online in Digital Health and Informatics Innovations for Sustainable Health Care Systems. Dr. Arlene Oetomo and Dr. Irfhana Zakir Hussain also contributed to this work as PhD candidates, with Zakir Hussain continuing this research at UbiLab.
