JMIR Publications

ABSTRACT

Background:

With the advancement of artificial intelligence technology, although robots have emerged as a new means of delivering health information, individuals still face challenges in deciding whether to trust the health information provided by these robots due to various trust-related factors.

Objective:

The aims of this study were to investigate the factors associated with the level of trust in robots providing health information among the general population in China, from a socioecological model perspective and to identify the central indicators based on network analysis.

Methods:

A nationwide survey in China was conducted from June 20th to August 31st, 2023, involving 30054 participants. The level of trust in robots providing health information was measured using a self-developed questionnaire. We utilized random forest classification to identify potential factors associated with the level of trust in robots providing health information grounded in the socioecological model. We also performed univariate and multivariable generalized linear model analysis to investigate the factors associated with the level of trust in robots providing health information. Network analyses were conducted to examine the network structure of trust levels in robots providing health information and associated factors.

Results:

The results of the multivariate generalized linear model analysis revealed significant positive associations between higher levels of self-rated health status (β = 0.15; 95% CI = 0.13 to 0.17), family health (β = 0.09; 95% CI = 0.03 to 0.14), perceived social support (β = 0.47; 95% CI = 0.39 to 0.55), agreeableness (β = 0.36; 95% CI = 0.15 to 0.57), openness (β = 0.49; 95% CI = 0.31 to 0.67), medical insurance type [commercial insurance and socialized medicine (β = 2.29; 95% CI = 0.80 to 3.77)], number of house properties [1 (β = 2.06; 95% CI = 1.13 to 2.99), ≥2 (β = 2.94; 95% CI = 1.96 to 3.92)] with the level of trust in robots providing health information. However, participants who exhibited the personality trait of neuroticism (β = -0.35; 95% CI = -0.55 to -0.15) and those of older age (β = -0.05; 95% CI = -0.07 to -0.03) demonstrated a significant negative association with the level of trust in robots providing health information. In addition, using network approach, central indicators were identified in the network of the level of trust in robots providing health information and its associated factors, including family health and perceived social support.

Conclusions:

Our findings not only offer a novel perspective on the association between robots providing health information and trust but also contribute to the application and development of artificial intelligence information technology. By considering factors associated with the level of trust in robots providing health information, the acceptance and adherence of individuals to health information may be enhanced.