Trustworthiness and relevance in web-based clinical question answering

Area of Science:

• Medical Informatics
• Natural Language Processing
• Information Retrieval

Background:

• Clinical question answering (QA) systems aim to provide precise and relevant answers to user queries.
• Ensuring the trustworthiness of answers is critical in the clinical domain.
• Limiting document sources to certified websites enhances trustworthiness but may reduce answer relevance.

Purpose of the Study:

• To evaluate the impact of prioritizing trustworthiness on answer relevance in a clinical QA system.
• To determine if limiting document sources to certified websites negatively affects the relevance of retrieved answers.

Main Methods:

• A comparative evaluation of a clinical QA system was conducted.
• The system's document collection was limited to certified websites to enhance trustworthiness.
• Answer relevance and trustworthiness were assessed.

Main Results:

• Promoting trustworthiness by using certified websites did not negatively impact answer relevance.
• The clinical QA system demonstrated improved answer relevance when using trustworthy sources.
• Both trustworthiness and relevance of answers were enhanced.

Conclusions:

• Limiting document sources to certified websites is an effective strategy for clinical QA systems.
• Enhancing trustworthiness in clinical QA does not compromise, and can even improve, answer relevance.
• This approach supports the development of more reliable and effective clinical decision support tools.