A recent study from UC Irvine suggests that social media conversations could serve as a complementary tool for understanding how communities experience extreme heat and what specific aid they might need.
Researchers from the Joe C. Wen School of Population & Public Health, the Department of Statistics, and the Department of Computer Science analyzed 24,598 heat-related posts on Twitter, now X, from across California. The posts were collected between 2016 and 2022, providing a broad view of public discussion during periods of high temperatures.
The study found that a person’s residential situation, their methods of travel, and available local resources were associated with distinct ways they discussed extreme heat. This included whether they detailed the heat’s effects on their daily routines or focused on coping mechanisms. Suellen Hopfer, an associate professor of health, society and behavior and a senior author of the study, noted that social media gives "another window into how people are experiencing heat in their daily lives." She added that understanding these experiences can help refine heat-safety messages, identify communication gaps, and better support vulnerable communities.
As extreme heat becomes more frequent and intense across California, identifying which communities are struggling and their specific needs is increasingly important for public health. The UC Irvine team's approach went beyond tracking temperatures or heat-related illnesses; they examined how individuals described the daily impacts of heat and their coping strategies.
To conduct their analysis, researchers categorized posts by whether they discussed coping strategies—such as staying hydrated, using air conditioning, or changing exercise times—or life impacts, which included health effects, disrupted work, canceled activities, and concerns for pets or plants. These patterns were then compared with county-level data on income, housing, transportation, and tree canopy. A large language model was employed to group thousands of posts, with its accuracy verified against manually sorted posts, showing a range of 74 to 90 percent accuracy across different categories.
One of the clearest distinctions emerged regarding housing. Counties with a high prevalence of households living in RVs, vans, or boats, categorized as 'sensitive housing,' were found to be 15 percent less likely to post positive heat-related messages than counties with fewer such homes. These same communities had 22 percent lower odds of posting about coping strategies. Researchers highlighted the significance of this finding, noting that such residences often have poor insulation and can become dangerously hot, suggesting a need for targeted public health messages focused on indoor heat safety for these residents.
Transportation patterns also revealed specific vulnerabilities. Counties with higher percentages of residents who commute by walking, biking, or using public transportation exhibited a greater likelihood of posting negative heat-related comments and descriptions of how heat affected their daily lives. Researchers explained that active commuters face direct heat exposure during travel, indicating that improved transportation infrastructure could play a role in protecting these communities from rising temperatures.
The study also identified disparities in counties with larger Hispanic populations, which showed a slightly greater likelihood of posts describing heat’s impacts on daily life and a lower likelihood of posts on coping strategies. While the researchers cautioned against treating social media patterns as direct measures of population-level vulnerability, they emphasized that these results point to opportunities for more focused heat awareness and intervention efforts.
The UC Irvine researchers view social media not as a replacement for established public health surveillance methods, but as a supplementary source of information. Online conversations, which capture immediate experiences like struggling with an overheated home or altering a commute, can reveal aspects of heat exposure difficult to gather through conventional datasets alone. The study observed that heat-related posting increased in line with rising average monthly temperatures.
By combining these social media insights with existing data on housing, transportation, and socioeconomic conditions, public health agencies could eventually identify where particular kinds of heat communication are most needed. For instance, communities with significant sensitive housing could benefit from messages about maintaining safe indoor temperatures. Areas with many active commuters might benefit from heat-conscious transportation planning and interventions to reduce exposure during travel.
Ultimately, the study suggests that integrating social media monitoring with automated language analysis could provide an additional method for identifying communities that would benefit most from targeted heat warnings, practical guidance, and long-term mitigation strategies. As California anticipates more frequent and severe heat events, these findings could help public health officials move away from generic warnings toward communication tailored to how people genuinely experience and cope with extreme heat.
