Improving Access to Online Health Information With Conversational Agents: A Randomized Controlled Experiment

Area of Science:

• Health Informatics
• Human-Computer Interaction
• Clinical Trial Recruitment

Background:

• Conventional web search engines present barriers for individuals with low health literacy seeking health information.
• This limitation can exclude a significant population from accessing crucial health resources and research opportunities.

Purpose of the Study:

• To evaluate a conversational search engine interface designed to assist individuals with low health and computer literacy in finding clinical trials.
• To compare the usability and effectiveness of a conversational interface against traditional keyword-based search engines for clinical trial discovery.

Main Methods:

• A randomized trial with 89 participants compared a conversational search engine interface with a conventional keyword- and facet-based interface for the National Cancer Institute Clinical Trials database.
• Participants completed two tasks: finding a clinical trial for themselves and finding a trial meeting specific criteria.

Main Results:

• Participants reported significantly higher satisfaction with the conversational interface for both tasks compared to the conventional interface.
• The conversational interface led to a higher success rate in finding trials that met prespecified criteria, particularly for participants with low health literacy (36% success vs. 0% with conventional interface).

Conclusions:

• Conversational agents can substantially improve the accessibility of web-based searches, specifically for clinical trial information.
• This approach can help mitigate recruitment bias and increase participation from disadvantaged populations in clinical research.