Explainable online health information truthfulness in Consumer Health Search

Area of Science:

• Information Science
• Health Informatics
• Computer Science

Background:

• Online health information significantly impacts public health decisions.
• Current systems often use binary classification (true/false) for health information, lacking interpretability.
• Existing methods present opaque results, hindering user trust and decision-making.

Purpose of the Study:

• To develop a novel system for assessing the truthfulness of online health information.
• To address the limitations of binary classification and opaque results in current health information assessment tools.
• To enhance the interpretability of health information retrieval for consumers.

Main Methods:

• The study reframes health information assessment as an ad hoc retrieval task, not a classification task.
• An Information Retrieval model incorporating truthfulness as a relevance dimension was extended.
• Explainability was achieved by integrating a knowledge base of scientific evidence from medical journals.

Main Results:

• The proposed solution was evaluated quantitatively and qualitatively through a user study.
• Results demonstrated the system's effectiveness in providing interpretable ranked lists of documents.
• The system improved user understanding of both topical relevance and truthfulness of health information.

Conclusions:

• The novel approach enhances the interpretability of online health information for consumers.
• The explainable retrieval system aids users in making more informed health decisions.
• This work contributes to more trustworthy and transparent online health information access.