IDENTIFYING DETERMINANTS OF VACCINE HESITANCY FROM TWITTER DATA

Abstract

Background: Vaccine hesitancy refers to the delay of vaccination despite vaccine availability and recommendations from authorities. Conventional methodologies for assessing public attitudes toward vaccination, such as surveys and interviews, provide valuable but limited insights due to their static nature and inability to reflect real-time sentiment shifts. In comparison, Twitter data provide a broader and more dynamic perspective on public health behavior. Objectives: The thesis aims to determine whether mining Twitter data can complement conventional techniques to identify determinants of vaccine hesitancy. Methods: To achieve the goal, a scoping review was performed to identify determinants of vaccine hesitancy reported in the literature. A thorough survey to gauge public sentiments towards vaccines in the Middle East and North Africa (MENA) region was lacking. For this reason, the MENA region was surveyed to determine vaccine hesitancy and its determinants. After that, Twitter data were downloaded and analyzed. Topic modeling was performed to identify Twitter topics. Next, the topics were mapped to behavioral models to distinguish determinants of vaccine hesitancy. Lastly, vaccine hesitancy was measured through Twitter user stance analysis. Finally, results obtained from Twitter were compared to the review and survey. Results: Scoping review and MENA region survey yielded 19 and 11 determinants of vaccine hesitancy, respectively, while Twitter data mining revealed 25 determinants, many overlapping with conventional methods. Also, the comparative measures of vaccine hesitancy between Twitter data analysis and conventional surveys showed a similar response. For instance, Twitter analysis revealed hesitancy rates of 25.46% in Kuwait and 21.37% in Saudi Arabia, closely mirroring survey findings of 25.6% and 28%. iv Conclusion: This thesis contributes to the vaccine hesitancy literature by demonstrating the potential of using Twitter data as a complement to the survey methods to determine the determinants of vaccine hesitancy. It concludes with the recommendation of a multi-faceted approach, combining conventional survey methods and Twitter data analysis techniques for a comprehensive understanding of complex vaccine hesitancy issues.

Date of Award	2024
Original language	American English
Awarding Institution	HBKU College of Science and Engineering