Seasonal allergies, which affect around 26% of Americans, come with significant economic and productivity costs, especially as climate change drives higher pollen and mould levels. Traditional monitoring methods, including patient visits and pollen tracking, have limitations, such as low spatiotemporal resolution and inadequate capture of population-level trends. These gaps highlight the need for innovative approaches, prompting researchers to explore internet-based surveillance to track allergy trends in real-time across the U.S.
Internet-based surveillance, widely applied to track diseases like influenza and COVID-19, can potentially map trends in allergic diseases. Platforms like Twitter, Google, and Facebook provide high-volume data that reflect public health trends through text analysis and user search patterns. However, these methods face challenges; for instance, Google Flu Trends once failed to predict flu surges, sparking debate over accuracy and prompting integration with traditional surveillance data to improve reliability.
In this study, researchers used historical search data from Google Trends and Twitter to quantify allergy-related patterns, validated against emergency department (ED) visits coded for allergic conditions in California from 2016 to 2019. Machine learning identified correlations between online activity and hospital visits, finding a statistically significant relationship. This finding allowed researchers to model allergy trends at daily resolutions across 144 U.S. counties, providing high spatiotemporal accuracy over an 8-year period.
The model highlights regional and temporal variations in allergy patterns, offering potential for early detection. For instance, the methodology detected a sharp allergy increase in Los Angeles County in June 2018, warranting further exploration of regional aeroallergen drivers, such as mould or specific pollens, through ground-based measurements.
This multi-source approach combines internet-based and traditional surveillance for more robust allergy tracking. While effective in urban areas, this method requires a population base large enough for reliable data. Future studies could address this limitation through semantic analyses and custom classifiers for improved predictions in sparsely populated regions. With further refinement, internet-based allergy tracking could complement real-time aeroallergen monitoring, informing public health efforts and enabling timely, geographically targeted responses to seasonal allergy trends across the U.S.
Reference
Stallard-Olivera E, Fierer N. Internet-based surveillance to track trends in seasonal allergies across the United States. PNAS Nexus. 2024;3(10):pgae430.
