No abstract

No abstract

No abstract

Background: The COVID-19 pandemic is an awakening call for public health agencies. Digital technologies create a decentralized information environment in which public health agencies must compete for audience attention and win their trust. Trust is a result of inclusiveness of all stakeholders, mutual understanding, and recognition of different perspectives. Thereby, public health agencies should apply two-way communication and cognitive empathy, by listening to local communities. Technology advancement in Artificial Intelligence has made it possible to “listen” to many stakeholders on social media. This study urges a focus on listening at local levels, for example, cities, given the abundance of geo-marked data, and the importance of community-level operations to manage public health crises.Methods: The case study presented combined AI methods with textual analysis and examined 180,128 tweets posted by four cities with large populations of people of color. Results: The findings discovered sentiment around “COVID Vaccines,” “Politics,” “Mitigation Measures,” and “Community/Local Issues” and critical moments of emotional changes.Conclusions: Our major contribution is to explain the motivation and the methods of extracting intelligence for the purpose of enhancing public trust in health agencies during crises.

BACKGROUND COVID-19 vaccination rates have waned across the country since the rollout in early 2021, especially among African American neighborhoods. Vaccine hesitancy is a recurring theme challenging the world’s public health. Yet, months after efforts to vaccinate the world’s population, we still do not have a good understanding of consumer insights about those who choose to be vaccinated and those who refuse. This also suggests that many vaccination campaigns are running on assumptions, not evidence-informed by consumer insights. OBJECTIVE The purpose of this study is to understand consumer insight of COVID-19 vaccines in Kansas City, a city with higher percentages of African Americans, to contextualize the insight and further compare data from Kansas City with insights from three other similar-sized towns (Long Beach, California; Omaha, Nebraska; Raleigh, North Carolina) that also have higher percentages of people of color. METHODS The researchers collected and analyzed 180,128 tweets from four cities. Triangulated methods were used to look at the breadth and depth of data to provide validity to the findings. In addition, health communication experts, informed by machine learning/deep learning topic and emotion models, conducted a textual analysis of the tweets. The strength of this study is the compilation of methods and the ways in which the data was analyzed and visualized. RESULTS Four major themes about COVID were discovered from the mass of tweets: “COVID Vaccines,” “Politics,” “Mitigation Measures,” and “Community/Local Issues.” The tweets per topic and emotion category were visualized to show regional differences and longitudinal changes. Critical moments of emotional changes were detected. Textual analysis based upon data partitioned by the models identified national and local themes. Insights into strategies of appealing to residents are discussed. CONCLUSIONS This project’s data reveal wavering relationships of trust among residents and the government and its entities. While long-term initiatives should be used to rebuild and strengthen relationships among residents in cities with higher percentages of people of color, additional attention should be given to the health messaging directed at this audience. Practical implications are offered to inform local vaccination campaigns.