Reluctance or refusal to get vaccinated, referred to as vaccine hesitancy (VH), has hindered the efforts of COVID-19 vaccination campaigns. It is important to understand what factors impact VH behavior. This information can help design public health interventions that could potentially increase vaccine uptake. We develop a random forest (RF) classification model that uses a wide variety of data to determine what factors affected VH at the county level during 2021. We consider static factors (such as gender, race, political affiliation, etc.) and dynamic factors (such as Google searches, social media postings, Stringency Index, etc.). Our model found political affiliation and the number of Google searches to be the most relevant factors in determining VH behavior. The RF classification model grouped counties of the U.S. into 5 clusters. VH is lowest in cluster 1 and highest in cluster 5. Most of the people who live in cluster 1 are democrat, are more internet-inquisitive (are more prone to seek information from multiple sources on the internet), have the longest life expectancy, have a college degree, have the highest income per capita, live in metropolitan areas. Most people who live in cluster 5 are republicans, are the least internet-inquisitive, have the shortest life expectancy, do not have a college degree, have the lowest income per capita, and live in non-metropolitan areas. Our model found that counties in cluster 1 were most responsive to vaccination-related policies and COVID-19 restrictions. These strategies did not have an impact on the VH of counties in cluster 5.
翻译:我们开发了一个随机森林分类模式,使用各种数据来确定2021年期间县一级哪些因素影响县一级健康,称为疫苗失常(VH),阻碍了COVID-19疫苗接种运动的努力。重要的是要了解哪些因素影响VH行为。这种信息有助于设计可能增加疫苗摄入量的公共卫生干预措施。我们开发了一个随机森林分类模式,使用各种数据来确定2021年期间在县一级影响VH的因素。我们考虑的是静态因素(如性别、种族、政治派别等)和动态因素(如谷歌搜索、社交媒体张贴、严格指数等)。我们的模型发现政治归属和谷歌搜索数量是确定VH行为的最相关因素。我们发现的政治归属和谷歌搜索数量是确定VH行为的最相关因素。RF分类模式将美国各州分组分为5组。 VH在群组中最低,在群组中居于第1组的人大多为不民主,互联网上的多数人(更容易从多种来源获取信息)和动态因素(如谷歌搜索、社交媒体张贴、谷歌搜索数量最长的大学学位、人均收入最高、在城市一级有最低收入限制的地区。