Computational Intelligence-Based Model for Exploring Individual Perception on SARS-CoV-2 Vaccine in Saudi Arabia

Area of Science:

• Public Health
• Computational Linguistics
• Epidemiology

Background:

• The global challenge of controlling SARS-CoV-2 spread necessitates effective vaccination strategies.
• Over 50 SARS-CoV-2 vaccine candidates are in development, with limited deployment.

Purpose of the Study:

• To analyze public awareness and opinion regarding SARS-CoV-2 vaccination in Saudi Arabia.
• To develop a predictive model for vaccine awareness and acceptability using social media data.
• To inform public health campaigns and governmental strategies for vaccine promotion.

Main Methods:

• Analysis of Arabic tweets related to SARS-CoV-2 vaccination.
• Application of machine learning models including Support Vector Machine (SVM), Naïve Bayes (NB), and Logistic Regression (LR).
• Feature extraction using N-gram and Term Frequency-Inverse Document Frequency (TF-IDF).
• Implementation of a Long Short-Term Memory (LSTM) model with word embedding.

Main Results:

• Logistic Regression with unigram feature extraction achieved 0.76 accuracy, 0.69 recall, and 0.72 F1-score.
• Support Vector Machine (SVM) with unigram and Naïve Bayes (NB) with bigram TF-IDF yielded the highest precision (0.80).
• The Long Short-Term Memory (LSTM) model demonstrated superior performance with 0.95 accuracy, precision, recall, and F1-score.

Conclusions:

• The Long Short-Term Memory (LSTM) model effectively predicts public awareness and acceptability of SARS-CoV-2 vaccines.
• Understanding public sentiment through social media analysis is crucial for targeted health interventions.
• Findings can guide government initiatives to enhance vaccine awareness and acceptance in Saudi Arabia.