Introduction
The COVID-19 pandemic has changed the lives of individuals and societies worldwide. Since its identification by the WHO in late 2019, the USA alone has since seen more than 87 million confirmed cases and 1 million virus-related deaths.1 With millions of people forced out of public spaces, conversations surrounding the pandemic have taken place online, with a vast majority via social media networks.2 3 People in the Public Eye (PIPE) such as news anchors, politicians, athletes and entertainers have taken this opportunity to discuss diverse topics with their platform followers, including their own experiences and opinion about health issues3 such as COVID-19, COVID-19 vaccines and other public health measures. These shared experiences and opinions by PIPE often give way to public discourse, with many reacting to polarising statements by liking, commenting and resharing. As populations grow to trust the influential nature of celebrity activity on social platforms, followers are disarmed and open to persuasion when faced with false information, creating opportunities for dissemination and rapid spread of misinformation and disinformation.4 5 Exposure to large amounts of misinformation can have lasting effects on overall well-being and sociopsychological health.6 We argue this threat to population health should create a sense of urgency and warrants public health response to identify, develop and implement innovative mitigation strategies.
Background
As social media use has sustainably grown amid the COVID-19 pandemic, public health researchers have capitalised on the potential for data mining of shared messages on social platforms.7 Ease of access and rapid collection of data permits researchers to follow pandemic progression alongside online sentiment, providing tools for niche discovery and exploration of the emotion behind health decision making. For example, mining tweets from a specified period allows for parallel analysis of general public opinion during major events (ie, the release of new treatments such as vaccines or the death of a celebrity post-COVID-19 infection). Regarding COVID-19 vaccination specifically, researchers have used this recent increase in opinion sharing to measure overall sentiment and vaccine hesitancy or acceptance.8–11 Moreover, many studies published over the past two decades highlight the persuasive nature of celebrity behaviour and messaging—both beneficial and detrimental to public health.12–15 Others have used emotional diffusion networks to investigate the correlation between messaging shared by governmental agencies on social platforms and subsequent sentiment shared in response by the general public.16 All present strong evidence in support of impacts on health-related perception, emotion and behaviour as a result of sentiment shared by those with societal influence or authority. Sentiment analysis is the practice of extrapolating the sentiment of a subject, idea, event or phenomenon by classification of written texts as some value of polarity (ie, positive or negative).17 In our previous works,18–20 our team has successfully employed various natural language processing (NLP) models for the analysis of social media shared sentiment. The application of such analytic tools could allow for time-expanded retrospective analysis of online sentiment in correlation with the progression of the ongoing pandemic. This targeted approach could provide tools for niche discovery and exploration of the emotions behind health decision making throughout the COVID-19 pandemic era and enhance preparedness, response and recovery efforts for future health crises.
Herein, we argue that, with the utilisation of a fine-tuned DistilRoBERTa NLP model,18 sentiment and content analysis could uncover a correlation between COVID-19 vaccine-related messaging shared by PIPE and public sentiment and discourse direction. This discovery could aid in better understanding public perception and attitude towards vaccination based on social influences, providing officials and policymakers tools to combat mis/disinformation shared via social media platforms moving forward.