Acceptance of healthcare services based on the large language model in China: a national cross-sectional study

Increasing public acceptance of medical large language models will be beneficial for further leveraging their potential in reducing medical costs and improving efficiency. The objective of our research is to figure out the acceptance of healthcare services based on large language models in China, and to determine the demographic characteristics and related cognitive factors associated with it.

This cross-sectional study was conducted in 31 provinces in mainland China through an online survey using the China network questionnaire platform (Wen Juan Xing). The data was collected from April 21 to May 13, 2025. The report analysis period was from May to August 2025. We initially set the sample size at 3,000 people, and rounded up based on the preliminary calculation results. The proportion of questionnaires distributed in each province was the same as the proportion of the reported population in each province. A total of 3,148 valid questionnaires were ultimately collected. The main outcome was the acceptance of medical large language models. Univariable and multivariable logistic regression models were performed to explore the associations between individual factors and the acceptance.

Among 3,148 Chinese residents, 57.9% (95% CI: 56.22-59.66%) were willing to accept healthcare services based on large language models. In the multivariate logistic regression model, acceptance of large language models for healthcare services was significantly associated with being male (aOR = 1.216, 95% CI: 1.028-1.438), high awareness of artificial intelligence (aOR = 2.386, 95% CI: 1.911-2.980), and low concern about fairness (aOR = 2.233, 95% CI: 1.037-4.809) and privacy disclosure (aOR = 3.805, 95% CI: 2.139-6.767).

Increasing public exposure to large language models through formal and official channels may help enhance the acceptance of healthcare services based on large language models.