Improving medical experts' efficiency of misinformation detection: an exploratory study.

Area of Science:

• Medical Informatics
• Public Health
• Computational Linguistics

Background:

• The proliferation of medical disinformation, especially during pandemics, necessitates accurate credibility assessment systems.
• Current automated systems lack the precision for reliable medical information verification, requiring expert human annotators.
• Optimizing the time and effort of medical experts is crucial for efficient fact-checking.

Purpose of the Study:

• To develop a framework for semi-automatic verification of medical information credibility.
• To optimize the utilization of medical experts' time by focusing on non-credible statements.
• To enhance the efficiency of identifying medical disinformation.

Main Methods:

• Implementation of a general framework for filtering medical statements.
• Construction of specialized filtering classifiers adapted to narrow thematic medical categories.
• Quantitative and exploratory analysis of filtering classifier performance across diverse medical topics.

Main Results:

• Medical experts identified over two times more non-credible medical statements per unit of time.
• The framework successfully filtered out medical statements not requiring manual expert evaluation.
• The approach demonstrated consistent performance across a broad spectrum of medical topic areas.

Conclusions:

• The proposed framework significantly enhances the efficiency of medical experts in fact-checking and identifying disinformation.
• Specialized filtering classifiers allow for a more focused and productive use of expert annotators' time.
• The system offers flexibility for modification to provide varied expert feedback without compromising performance.