JMIR Publications

ABSTRACT

Background:

Among the alternative solutions being tested to improve access to genetic services, chatbots (or conversational agents) have been increasingly used for service delivery. Despite the growing number of studies on the accessibility and feasibility of chatbot genetic service delivery, limited attention has been given to user interactions with chatbots in a real-world healthcare context.

Objective:

We examined users’ interaction patterns as well as the associations of users’ clinical and sociodemographic characteristics, chatbot interaction patterns, and genetic testing decisions with a cancer genetics pretest education chatbot.

Methods:

We analyzed data from the experimental arm of Broadening the Reach, Impact, and Delivery of Genetic Services (BRIDGE), a multi-site genetic services pragmatic trial, in which participants eligible for hereditary cancer genetic testing based on family history were randomized to receive a chatbot intervention or standard of care. In the experimental chatbot arm, participants were offered access to core education content delivered by the chatbot with the option to select up to nine supplementary informational prompts and ask open-ended questions. We computed descriptive statistics on the following interaction patterns: prompt selections, open-ended questions, completion status, dropout points, and post-chat genetic testing decisions. Logistic regression models estimated the relationships of clinical and sociodemographic factors with chatbot interaction variables to examine how these factors affected genetic testing decisions.

Results:

Of 468 participants who started the chat, 391 (83.5%) completed it, with 315 (80.6%) of whom competed expressing willingness to pursue genetic testing. Among completers, 336 (85.9%) selected at least one informational prompt, 41 (10.5%) asked open-ended questions, and 0.7% opted for extra examples of risk information. For non-completers, 57 (74.0%) dropped out before accessing any informational content. Interaction patterns were generally not associated with clinical and sociodemographic factors, except for prompt selection (varied by study site) and completion status (varied by family cancer history type). Participants who selected three or more prompts (OR = 0.33; 95% CI: 0.12, 0.91; p = 0.03) or asked open-ended questions (OR = 0.46; 95% CI: 0.22, 0.96; p = 0.04) were less likely to opt for genetic testing.

Conclusions:

Our findings highlighted the chatbot’s effectiveness in engaging users and high acceptability, with most participants completing the chat, opting for additional information, and showing a high willingness to pursue genetic testing. Sociodemographic factors were not associated with interaction patterns, potentially indicating the chatbot’s scalability across diverse populations if they are able to access the chat. Future efforts should address high-information users’ concerns and integrate them into chatbot design to better support informed genetic decisions.