The COVID-19 pandemic accelerated the spread of misinformation online, creating an “infodemic” that had profound effects on health behavior [1]. The breadth and depth of COVID-19 misinformation expanded to include all vaccinations, such as human papillomavirus (HPV) vaccination, depressing already suboptimal vaccination uptake in the United States [1,2]. As HPV vaccination is critical to the prevention of various cancers, this could pose significant cancer control challenges in the future [2]. There is an urgent need to address HPV vaccination misinformation to increase HPV vaccination uptake [2]. Behavioral interventions can counter misinformation online, but they are typically limited to a single social media platform without geographic specificity [3].

Public health surveillance (PHS) is defined by the Centers for Disease Control and Prevention (CDC) as “the ongoing, systematic collection, analysis, interpretation, and dissemination of data regarding a health-related event for use in public health action to reduce morbidity and mortality and to improve health” [4]. Digital PHS (DPHS) uses data from online sources, often collected outside of traditional PHS, for similar purposes [5]. There has been debate as to the ethics of using publicly available online data for DPHS [5]. However, the pandemic illustrated the need for user-friendly, timely, interactive digital tools to drive health-related intervention [6].

Social listening (SL) is the process of aggregating data from across online channels to collect real-time measures of emotions, opinions, and themes, typically through platform algorithms that rely on machine learning and artificial intelligence (AI) [7]. While SL platforms’ AI-driven tools for emotion and sentiment detection can be unreliable, machine learning provides an opportunity to “train” SL platforms for greater accuracy over time in the automated recognition of emotions and sentiments [8]. The World Health Organization Early AI-Supported Response With Social Listening Platform (WHO EARS) uses an SL dashboard to provide health professionals access to information from across the internet to assist in the development of timely responses to COVID-19 narratives that occur online at the global and country levels, highlighting the growing acceptance of such tools in public health [7].

The purpose of this exploratory study was to assess the feasibility of using a commercially available SL platform to monitor HPV vaccination misinformation online at the national (ie, within the United States overall) and state (ie, within Mississippi and Rhode Island) levels.

Table 1 details each adapted PHS system attribute and the opportunities and limitations with regard to using a commercially available SL platform for HPV vaccination misinformation DPHS. Opportunities include user-friendly dashboards with real-time data monitoring and platform adaptability. For example, from June 21 to 24, 2023, the research team was able follow the spread of misinformation through social media posts related to a lawsuit filed by the Children’s Health Defense Fund, an organization led by prominent antivaccine activist Robert Kennedy Jr. However, while the SL platform dashboards are user-friendly, it took significant staff time, expertise, and routine maintenance to keep them relevant and as accurate as possible. Brandwatch was also found to be adaptable to the ever-changing online information ecosystem; however, the quality of this information was dependent on data access agreements with individual social media companies, which could change at any time. Additional challenges to using an SL platform for DPHS include concerns with data quality, representativeness, and the accuracy of AI-driven tools. There are limited ways to validate data within the SL platform itself. Data may be downloaded from Brandwatch and externally analyzed for sentiments and emotions, but this process would remove the AI-driven, automated nature of the SL platform and reduce the effectiveness of real-time monitoring in DPHS.

While Brandwatch was selected due to opportunities for greater geographic specificity, this functionality was limited in scope to only certain social media platforms, such as X (formerly Twitter). Furthermore, geographic specificity was limited based on whether social media users used geolocation functionalities and whether locations were mentioned in profiles or posts. Despite this, the research team identified and monitored different narratives in misinformation within the two states included in this exploratory study—Rhode Island and Mississippi—suggesting the potential importance of assessing online misinformation narratives based on geographic location. For example, on the same day in January 2024, the top trending story for Rhode Island focused on the Children’s Health Defense Fund lawsuit, while in Mississippi, the top story focused on childhood injury due to vaccination.

Our findings suggest that there are opportunities and challenges associated with using commercially available SL platforms to monitor HPV vaccination misinformation online at the national and state levels. While there were strengths across all PHS system attributes, there were also significant weaknesses. These weaknesses, particularly those related to data quality, representativeness, and the accuracy of AI-driven tools, reflect limitations to using current SL platforms for DPHS. If these challenges are addressed over time however, this level of DPHS could provide the foundation for different intervention opportunities, such as using skilled infodemiologists to counter online misinformation [11]. While the research team identified challenges with the accuracy of Brandwatch AI-driven tools, which matched previously published research [8], building DPHS capabilities now could provide critical infrastructure if and when such tools improve over time. If found to be effective in monitoring HPV vaccine misinformation, commercially available SL platforms may be adapted to other fields and health conditions. Findings may differ based on the SL platform used and vendor access agreements with social media companies. Future research should focus on increasing the specificity of geographic location, studying strategies to increase the accuracy of SL platform AI-driven tools, and testing targeted interventions using SL platforms.