Internet search queries as a proxy for pollen counts

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Wednesday, 7 January 2015: 2:15 PM
228AB (Phoenix Convention Center - West and North Buildings)
Matthew K. Parker, Mercer University, Macon, GA; and P. Schramm, S. Saha, A. Manangan, and G. Luber

Airborne pollen and mold spores are the primary cause for seasonal allergic rhinitis, affecting between 30-60 million people per year in the United States. Over the past several decades, the amount of pollen produced by plants such as oak and ragweed has been increasing. This trend has also been accompanied by lengthier and earlier allergenic pollen seasons. Although allergic rhinitis has a large public health impact, many areas in the United States either lack adequate pollen monitoring equipment or the staff and expertise to collect pollen counts. Because of this lack of pollen surveillance, we sought a proxy measure capable of accurately predicting pollen counts and seasonality. We analyzed the utility of Google search engine query data as a proxy measure for pollen counts. In previous studies, this search data has been found to be useful in predicting influenza epidemics, Lyme disease, and other seasonal health outcomes. We examined search volume indexes in the Atlanta metropolitan area for pollen and health related search terms such as “allergies,” “runny nose,” and brand-name over-the-counter anti-histamine medications. A comparison of Google search volume index and measured speciated Atlanta pollen data from 2004-2011 suggests that internet search engine query data could serve as an effective proxy measure for pollen counts. This would allow for areas without adequate pollen monitoring stations to predict the start and end of the spring and fall pollen seasons on the basis of historical trends of search engine query data related to allergic rhinitis. Since weekly Google Trends data is publicly available and geographically specific, using it as a proxy measure for pollen has significant implications. Specifically, the breadth of search engine data could be used to determine the relationship between pollen counts and meteorological conditions (e.g., temperatures, precipitation, humidity) at distinct geographic locations to develop a weather-based predictive model for pollen concentrations. Furthermore, the data could be used by allergists and other clinicians to help advise patients on when to start allergy medications to prevent the onset of allergic rhinitis and the exacerbation of other respiratory illnesses such as asthma.