Human papillomavirus (HPV) is the most common sexually transmitted infection in the United States, with over 30,000 new HPV-related-cancers are diagnosed annually. Although HPV vaccines have been approved by the Food and Drug Administration (FDA) since 2006 and recommended for routine vaccination for school-age girls and boys, vaccination rates remain low. One reason that has contributed to low vaccination rates is incorrect ?risk perceptions? around HPV vaccines such as the high perceived risks of adverse events or side effects from the HPV vaccine. Incorrect risk perceptions are often rooted in the false information about HPV vaccines that people are exposed to in their daily life, including social media. The impact of social media on health information is substantial. Negative social-media HPV-vaccine information has been found to have an association with low vaccination coverage. Given the negative consequences of false information, there is a need to develop a robust and scalable way to detect false HPV-vaccine information before it propagates and negatively impacts behavior. The overarching goal of the proposed research is to build a model to identify false HPV-vaccine information on Twitter, demonstrate its impact on individual risk perceptions and measure its underlying mechanisms on risk perception formation. We propose a novel approach to leverage machine learning, natural language processing, network analysis, crowdsourcing/expert data annotation, psycholinguistic analysis and statistical modeling to investigate the false HPV-vaccine information collectively (in terms of its detection and propagation patterns) and individually (in terms of its impact and underlying cognitive mechanisms). Our study will first build a computational model to detect false HPV-vaccine information on Twitter. By modeling the domain-specific HPV- vaccine related text content, information-veracity related linguistic features, individual and collective user behaviors, and dissemination patterns, our model will be able to detect false HPV-vaccine information before it gets verified and spreads widely. We will then investigate the impact of false HPV-vaccine information on risk perceptions around HPV vaccination operationalized by natural language processing methods and a developed HPV-vaccine Risk Lexicon. We will further conduct psycholinguistic analysis on the false HPV-vaccine information and use statistical modeling to uncover the underlying mechanism of risk perceptions. Our study will make a critical and timely contribution to identifying the false HPV-vaccine information and its impact, which has the potential to be applied to other health topics. This proposed project will also address the National Cancer Institute priorities in promoting HPV vaccines and combating misinformation in cancer prevention and control.
The uptake of human papillomavirus (HPV) vaccine remains low in part because of incorrect perceptions of vaccination risks, which has been linked to the spread of false HPV-vaccine information. The proposed study seeks to build a computational model to detect false HPV-vaccine information on social media (Twitter) and determine its impact on risk perceptions of the HPV vaccine. The findings will provide important contributions to understand the impact of false health information on HPV vaccination behavior and could be expanded to other health topics.
