Dynamic Topic Modeling to Mine Themes and Evolution during the Initial COVID-19 Vaccine Rollout

An Open Access article published in the Health Behavior and Policy Review Journal.

Authors:

Ankita Agarwal
Dixit Bharatkumar Patel
Emily Burwell
William Romine
Tanvi Banerjee

Objective:

In this paper, we identify the topics in the form of themes being discussed on Twitter about the COVID-19 vaccine during the period of initial rollout of the vaccines and their evolution every month with the scientific advancement and major events on the COVID-19 vaccine timeline.

Methods:

We collected tweets from Twitter API over a period of 3 months from December 1, 2020 to February 28, 2021 using the keyword, ‘COVID-19 vaccine’ and implemented dynamic topic modeling to identify topics in the form of themes being discussed. We then visualized the evolution of these themes every month with the news events during that time.

Results:

We found that 8 themes were discussed on Twitter during the initial rollout of COVID-19 vaccines. Within each theme, there were some unique key words found in a particular month or new key words that emerged from the previous month. These themes evolved with the trending news during that time period.

Conclusions:

Public discussions and health behavior of people about COVID-19 vaccines across different themes evolved over time. There was a dynamic and temporal shift in the perception of people regarding these vaccines coinciding with the scientific advancement and news events around the development, distribution, and administration of these vaccines.

Source: Health Behavior and Policy Review
Publisher: Paris Scholar Publishing Ltd.
Article Link: https://doi.org/10.14485/HBPR.10.3.1