Evaluating the Credibility of Online Health Information from Medical and Computer Science Perspective: A Systematic Literature Review (Preprint)

Aleksandra Nabożny, Małgorzata Chlabicz, Adam Wierzbicki, Adam Jatowt, Dariusz Jemielniak

Abstract: Background: To combat online health misinformation effectively, bridging knowledge across computer science and medical sciences disciplines is required. Objective: This paper provides a comprehensive overview of research supporting the credibility evaluation of medical content from both perspectives. It aims to illuminate critical gaps between the two approaches and propose solutions to fill them. Methods: First, we reviewed existing toolkits and guidelines created by medical experts to assess the credibility and quality of Online Health Information (OHI). We reviewed n1=28 papers from the medical domain. Second, we examined n2=82 articles describing the efforts of computer scientists to assess OHI automatically or semi-automatically. We grouped those articles by their aim, algorithms, types of utilized datasets, and features extracted from sample OHI. Results: Among our most essential findings lies the conclusion that few recent computational studies leverage expert-developed medical credibility assessment tools in constructing datasets. Datasets have significant differences in annotation protocols, basic definitions, data structures, and more. Last but not least, many classification models focus solely on textual rather than multimodal OHI features, despite the rise of image and video misinformation. Conclusions: This survey highlights the urgent need to integrate credibility evaluation techniques from medicine into computational pipelines and unify methodologies for future classification experiments. Additionally, the survey provides a foundation to guide future cross-disciplinary efforts combining medical expertise with AI scalability.

External IDs:doi:10.2196/preprints.72751