Predicting Norovirus in England Using Existing and Emerging Syndromic Data: Infodemiology Study

Area of Science:

• Epidemiology
• Public Health Surveillance
• Data Science

Background:

• Norovirus causes a significant global burden of gastroenteritis, affecting all ages.
• Currently, no licensed vaccine or antiviral treatment exists for norovirus.
• Effective early warning systems are crucial for nonpharmaceutical prevention and control strategies.

Purpose of the Study:

• To evaluate the predictive power of syndromic surveillance and emerging data sources for norovirus activity in England.
• To assess the utility of internet searches and Wikipedia page views in predicting norovirus trends across different age groups.

Main Methods:

• Utilized existing syndromic surveillance and emerging data (internet searches, Wikipedia views).
• Employed Granger causality to assess temporal precedence of variables against norovirus laboratory reports.
• Applied random forest modeling to determine variable importance based on mean square error and node purity.

Main Results:

• Syndromic surveillance data proved valuable in predicting norovirus laboratory reports in England.
• Wikipedia page views offered limited additional predictive improvement over Google Trends and existing syndromic data.
• Predictive relevance varied significantly across age groups and geographic regions, with some models explaining up to 60% of variance.

Conclusions:

• Existing and emerging data can aid in predicting norovirus activity in specific English regions and age groups.
• Vomiting, gastroenteritis symptoms, and historical search terms like 'stomach flu' were significant predictors.
• Variations in public health practices and information-seeking behaviors influence predictor relevance; internet search data offers insights into public understanding for communication strategies.