COVID-19 vaccine rejection causes based on Twitter people's opinions analysis using deep learning

Area of Science:

• Computational linguistics
• Public health informatics
• Social media analytics

Background:

• Vaccine hesitancy poses a significant global health threat, impacting vaccine utilization despite availability.
• Understanding public perceptions of COVID-19 vaccines is essential for developing effective public health strategies and improving vaccine acceptance.

Purpose of the Study:

• To analyze public sentiment towards COVID-19 vaccines using social media data.
• To identify the primary causes of vaccine rejection and hesitancy.
• To classify negative sentiments based on identified rejection factors.

Main Methods:

• Multi-class sentiment analysis was performed on tweets related to COVID-19 vaccines.
• Machine learning (ML) and deep learning (DL) models, including Decision Trees (DT), Gated Recurrent Unit (GRU), Logistic Regression (LR), and Long Short-Term Memory (LSTM), were employed.
• Latent Dirichlet Allocation (LDA) was used to identify key themes and causes of vaccine rejection from negative tweets.

Main Results:

• Sentiment analysis revealed Decision Trees (DT) achieved 92.26% accuracy and Gated Recurrent Unit (GRU) achieved 96.83% accuracy.
• Five major causes for vaccine rejection were identified: Lack of safety, Side effects, Production problems, Fake news/Misinformation, and Cost.
• For classifying rejection causes, Logistic Regression (LR) reached 89.97% accuracy, and Long Short-Term Memory (LSTM) achieved 91.66% accuracy.

Conclusions:

• Machine learning and deep learning models effectively analyze public sentiment on COVID-19 vaccines from social media.
• Identifying specific rejection factors like safety concerns and misinformation is critical for targeted public health interventions.
• The findings can guide vaccine manufacturers in refining marketing strategies to address public concerns and boost vaccine confidence.