Background: Twitter is a leading microblogging platform, with over 126 million daily active users as of 2019, which allows for large-scale analysis of tweets related to migraine. June 2020 encompassed the National Migraine and Headache Awareness Month in the United States and the American Headache Society’s virtual annual conference, which offer opportunities for us to study online migraine advocacy. Objective: We aim to study the content of individual tweets about migraine, as well as study patterns of other topics that were discussed in those tweets. In addition, we aim to study the sources of information that people reference within their tweets. Thirdly, we want to study how online awareness and advocacy movements shape these conversations about migraine. Methods: We designed a Twitter robot that records all unique public tweets containing the word “migraine” from May 8th, 2020 to June 23rd, 2020, within a 400 km radius of New Brunswick, New Jersey, United States. We built two network analysis models, one for the months of May 2020 and June 2020. The model for the month of May served as a control group for the model for the month of June, the Migraine Awareness Month. Our network model was developed with the following rule: if two hashtag topics co-exist in a single tweet, they are considered nodes connected by an edge in our network model. We then determine the top 30 most important hashtags in the month of May and June through applications of degree, between-ness, and closeness centrality. We also generated highly connected subgraphs (HCS) to categorize clusters of conversations within each of our models. Finally, we tally the websites referenced by these tweets during each month and categorized these websites according to the HCS subgroups. Results: Migraine advocacy related tweets are more popular in June when compared to May as judged by degree and closeness centrality measurements. They remained unchanged when judged by between-ness centralities. The HCS algorithm categorizes the hashtags into a large single dominant conversation in both months. In each of the months, advocacy related hashtags are apart of each of the dominant conversation. There are more hashtag topics as well as more unique websites referenced in the dominant conversation in June than in May. In addition, there are many smaller subgroups of migraine-related hashtags, and in each of these subgroups, there are a maximum of two websites referenced. Conclusion: We find a network analysis approach to be fruitful in the area of migraine social media research. Migraine advocacy tweets on Twitter not only rise in popularity during migraine awareness month but also may potentially bring in more diverse sources of online references into the Twitter migraine conversation. The smaller subgroups we identified suggest that there are marginalized conversations referencing a limited number of websites, creating a possibility of an “echo chamber” phenomenon. These subgroups provide an opportunity for targeted migraine advocacy. Our study therefore highlights the success as well as potential opportunities for social media advocacy on Twitter.
The Influence of New Surveillance Data on Predictive Species Distribution Modeling of Aedes aegypti and Aedes albopictus in the United States