PT60 - TRUST PREDICTS ADOPTION OF AI CHATBOT AS A VIRTUAL NUTRITION ASSISTANT AMONG HEALTHCARE PROFESSIONALS AND STUDENTS: A PATH ANALYSIS

Linked sessions

PT60

TRUST PREDICTS ADOPTION OF AI CHATBOT AS A VIRTUAL NUTRITION ASSISTANT AMONG HEALTHCARE PROFESSIONALS AND STUDENTS: A PATH ANALYSIS

Y. N. Hoang1,*, J.-S. Chang1,2,3,4,5

1School of Nutrition and Health Sciences, College of Nutrition, 2Graduate Institute of Metabolism and Obesity Sciences, College of Nutrition, Taipei Medical University, 3Nutrition Research Center, Taipei Medical University Hospital, 4Chinese Taipei Society for the Study of Obesity (CTSSO), 5TMU Research Center for Digestive Medicine, Taipei, Taiwan, Province of China

 

Rationale: While ChatGPT has shown promise as a virtual nutritionist, factors influencing its adoption and use remain unexplored. This study examined factors influencing the actual use of ChatGPT for nutrition-related academic and clinical tasks among healthcare professionals and students, with a focus on the role of trust.

Methods: A web-based survey was conducted among healthcare professionals and students in Taiwan, including dietitians, and future doctors and dietitians. The questionnaire was based on an extended Technology Acceptance Model (TAM), assessing perceived usefulness (PU), perceived ease of use (PEU), trust (TR), perceived risk (PR), attitude toward use (ATU), behavioral intention to use (BIU), and actual use (AU). Structural relationships were analyzed using partial least squares structural equation modeling (PLS-SEM).

Results: A total of 161 respondents (81.3% female; mean age = 21.4 ± 3.67 years) completed the survey. AU had the lowest mean score among constructs. ChatGPT was most commonly used for understanding nutrition knowledge (M = 3.27, SD = 1.16) and analyzing dietary records (M = 3.11, SD = 1.17); lower usage was reported for exam preparation (M = 2.92, SD = 1.13) and clinical dietary assessment (M = 2.86, SD = 1.13). The model explained 28.0% and 42.2% of the variance in AU for academic and clinical tasks, respectively. TR and BIU significantly predicted AU for academic tasks (e.g., understanding nutrition: TR β = 0.411; BIU β = 0.243; both p < 0.005). For clinical tasks, TR was the only significant predictor (e.g., dietary analysis: β = 0.534, p < 0.001).

Conclusion: Trust plays a critical role in the adoption of ChatGPT for both academic and clinical nutrition tasks. Enhancing trust in AI tools may facilitate their integration into nutrition care and education among healthcare providers and students.

Disclosure of Interest: None declared