Monitoring and Explaining Vaccine Refusal Using the Internet and Social Media

Program Delivery and Evaluation Network

Vaccination refusal on non-medical grounds is increasing in some developed countries, and areas with high rates of refusal tend to cluster geographically. Individuals who refuse vaccination are at an increased risk of acquiring and transmitting vaccine-preventable diseases, and local increases in refusal are associated with community outbreaks of disease. In general, people who refuse vaccination do so due to a belief that vaccines are not safe and that the risks from infection are low; the reasons for these beliefs vary according to factors such as gender, culture, and education. To be successful, public health interventions to improve vaccine coverage must change these beliefs by addressing the reasons for those beliefs in a specific jurisdiction. Increasingly, people are expressing their thoughts and beliefs about vaccination publicly on the Internet (e.g., blogs) or through social media (e.g., Twitter) and it is often possible to determine the geographic region (e.g., province, city, or even neighbourhood) of individual posts or tweets. We propose to use methods from artificial intelligence to automatically extract concepts from digital media (natural language processing), then organize the concepts using existing knowledge (knowledge modeling) to answer questions about reasons for vaccine refusal in specific sub-populations and geographic regions. This work will build on our previous research examining the determinants of vaccination and using artificial intelligence methods in surveillance, and will serve as pilot data for a future grant application. The results of this research should help public health practitioners to better understand the factors driving vaccine refusal, to identify potential interventions for increasing coverage, and to evaluate the effect of those interventions on perceptions and beliefs.