An applied study on the implementation of a multimodal explainable artificial intelligence system in long-term care institutions

This study investigates the effects of perceived ease of use (PEU), perceived usefulness (PU), and trust (TR) on behavioral intention (BI) toward adopting multimodal explainable artificial intelligence systems in long-term care institutions. Specifically, it examines the mediating role of PU in the relationship between PEU and BI and further analyzes how TR moderates the relationships of PEU and PU with BI. Data were collected through an online questionnaire administered to nurses, care attendants, and administrators working in long-term care facilities in Taiwan. A total of 150 questionnaires were distributed, and 125 valid responses were obtained, yielding an effective response rate of 84%. ncluded a clear, non-technical explanation of what a multimodal explainable AI system is, to help readers who are not AI specialists better understand the core conce. The empirical results indicate that PU, PEU, and TR all have significant positive effects on BI. Moreover, PU partially mediates the relationship between PEU and BI, while the interaction terms TR×PEU and TR×PU significantly moderate BI: when trust is high, the effect of PU on BI is strengthened, whereas the effect of PEU on BI is weakened. These findings suggest that, in high-trust settings, long-term care professionals are more willing to tolerate system complexity if the system is clearly useful. Based on these results, the study offers managerial and policy implications for integrating smart technologies into Taiwan’s long-term care sector, emphasizing the importance of usefulness, usability, and trust in enhancing technology adoption. This research extends the Technology Acceptance Model by integrating trust, providing new insights for digital transformation in long-term care