Detecting health misinformation: A comparative analysis of machine learning and graph convolutional networks in...

Area of Science:

• Digital Health
• Computational Linguistics
• Information Science

Background:

• The digital era presents significant challenges due to the widespread dissemination of health misinformation online.
• This misinformation poses a substantial threat to public health and well-being.
• Effective detection of health misinformation is critical for safeguarding public health.

Purpose of the Study:

• To conduct a comparative analysis of various classification models for detecting health misinformation.
• To evaluate the performance of traditional machine learning algorithms against advanced Graph Convolutional Networks (GCNs).
• To identify the most effective algorithmic approach for combating the spread of false health information.

Main Methods:

• A comparative analysis of classification algorithms was performed.
• Algorithms evaluated included Passive Aggressive Classifier, Random Forest, Decision Tree, Logistic Regression, Light GBM, GCN, GCN with BERT, GCN with TF-IDF, and GCN with Word2Vec.
• Performance was assessed using metrics such as accuracy, precision, recall, and F1-score.

Main Results:

• Graph Convolutional Networks (GCNs) combined with TF-IDF embedding demonstrated superior performance.
• GCN with TF-IDF achieved the highest accuracy at 93.86%.
• Other models showed varying levels of effectiveness, with Random Forest at 86% and Passive Aggressive Classifier at 85.75%.

Conclusions:

• Graph Convolutional Networks, particularly when utilizing TF-IDF embeddings, represent a highly effective method for detecting health misinformation.
• The findings provide valuable insights for developing robust systems to combat online health misinformation.
• This research highlights the potential of advanced network-based models in addressing critical public health information challenges.