Adoption and perception of LLM-based chatbots in health care: an exploratory cross-sectional survey of individuals with rheumatic diseases.

OBJECTIVE

The rapid mainstream uptake of artificial intelligence (AI) technologies, particularly large language model (LLM)-based chatbots, have sparked interest in their potential role in healthcare. Despite technological advancements, little is known about the current utilization of LLM chatbots among individuals with rheumatic diseases. This study aimed to investigate the adoption of and perceptions towards LLM chatbots among individuals with rheumatic disease, along with associated sociodemographic factors.

METHODS

An exploratory cross-sectional survey was conducted with participants recruited both online, via Arthritis Care Experts' digital and social media platforms, and in person from rheumatology clinics in Edmonton, AB, Canada. Respondents completed an 18-item questionnaire assessing LLM chatbot use for work and in daily life, including for health-related purposes, alongside sociodemographic factors. Chi-squared tests were used to assess crude associations and multivariable logistic regression was used to evaluate the adjusted odds ratios of sociodemographic factors and LLM chatbot use.

RESULTS

Of 270 respondents (109 online, 161 in person), 119 (44%) reported using LLM chatbots, with 40 respondents (15%) using them for health-related reasons. LLM chatbots were primarily used for general health queries rather than specific or personal health questions. Younger age and a more liberal political view were associated with LLM chatbot use, while gender, education, income, ethnocultural background and language spoken were not.

CONCLUSION

This study showed that a relevant number of individuals with rheumatic diseases are already using LLM chatbots, including for health-related reasons. These findings should prompt urgent efforts to address accuracy, safety and equity concerns regarding the utilization of LLM chatbots, particularly in the domain of rheumatology.