Populations, exposures and outcomes: How social media data is poised to advance climate-demography research

Abstract

As the natural environment continues to change, there is a shared sense among researchers, policymakers, and health practitioners that we need to better understand the impacts of those changes on health and health behaviors for individuals and populations. “Big” social media data, including large datasets from platforms like X (formerly Twitter), Reddit, Facebook, TikTok, and YouTube, provide access to previously hard-to-reach populations. These data are posed to revolutionize the data landscape in climate change research; however, there are limitations and ethical concerns with data from these platforms, such as ethical questions, biases, noise, and a distinct underrepresentation of marginalized groups. These issues could hamper the adoption of social media data in research on climate change and health. In this commentary, we propose leveraging and augmenting a “human-in-the-loop” approach that uses humans in data extraction, quality assurances, and model testing to ensure data quality and overcome known limitations of social media data. Using examples from health data, we propose new horizons in climate change health research made possible with qualitative research methods.

The authors' abstract, as published at the source. Proceedings of the National Academy of Sciences, 2026 · DOI ↗