Health Misinformation Detection: Approaches, Challenges and Opportunities

Area of Science:

• Digital Health
• Public Health Informatics
• Computational Social Science

Background:

• Health misinformation poses significant risks to public health.
• Effective detection of health misinformation is crucial for mitigation efforts.
• Existing research on detection methods requires comprehensive synthesis.

Purpose of the Study:

• To conduct a comprehensive literature review on health misinformation detection methods.
• To analyze the characteristics, datasets, and evaluation metrics of health misinformation.
• To examine the strengths and limitations of various detection approaches.

Main Methods:

• Systematic literature search of Google Scholar (Jan 2016 - Feb 2025).
• Inclusion of 100 full-text, English-language studies on health misinformation detection.
• Analysis of study characteristics, detection methods, datasets, and evaluation metrics.

Main Results:

• Machine learning and deep learning approaches show promise, with ensemble methods and embedding-based representations enhancing performance.
• Challenges include class imbalance, inconsistent annotations, high computational costs, and low interpretability of deep learning models.
• Advanced methods improve accuracy and explainability but introduce concerns regarding AI-generated misinformation and ethics.

Conclusions:

• The state-of-the-art in health misinformation detection requires interdisciplinary collaboration.
• Human-centered design and ethical considerations are vital for developing effective detection systems.
• Future research should address AI-generated misinformation and ethical implications.