This article is more than 5 years old. Information may no longer be current.

Twitter data may accurately predict ED visits for asthma

Social media platforms and environmental data may have the ability to accurately forecast ED visits for asthma at a population level, according to study results.

“Our population-level asthma risk prediction model has the potential for complementing current individual-level models, and may lead to a shorter time window and better accuracy of prediction,” Sudha Ram, PhD, of the University of Arizona, and colleagues wrote. “This in turn has implications for better planning and proactive treatment in specific geo-locations at specific time periods.”

The researchers built a preliminary prediction model to determine the accuracy of using multiple data sources to predict the number of asthma-related ED visits. The model was built using data streams from Twitter, air quality data obtained from sensors, as well as electronic health records and Google search trends.

The researchers obtained aggregate data on ED visits for asthma as a primary diagnosis at the Children’s Medical Center in Dallas from October to December 2013. Additional information was collected between November and December 2013 on ED visits for abdominal pain or constipation to serve as controls unrelated to asthma activity.

The Twitter dataset included 464,845,785 tweets, including 1,315,390 asthma-related tweets.

Google search interest data also were obtained to analyze the frequency of specific search terms for specific periods. Air pollution data were also drawn from Environmental Protection Agency databases.

The model predicted the number of asthma ED visits based on near-real-time environmental and social media data with about 70% accuracy. The researchers acknowledged, however, the data only came from the Children’s Medical Center. More research is being done in larger clinical samples in a larger geographic region.

“Clinical resources could be prioritized to offer earlier clinic appointments to patients with impending risk for failure, and later slots to patients with deferred risk,” the researchers wrote. “Additionally, hospitals and EDs could use such risk-stratification for optimal resource planning, such as ED staffing or opening observation units.”

Disclosure: The researchers report no relevant financial disclosures.