Page 431 - ISC PROCEEDINGS 21.4

< 0.001), supporting H5 and H8.
Mooring factors negatively affect Switching Intention and negatively moderate the
impact of Pull factors on the Switching Intention to AI-assisted healthcare (p < 0.001),
supporting H2 and H4.
The model explains 71.2% of the variance in Switching Intention (R² = 0.712),
indicating strong explanatory power. The remaining 28.8% of the variance in the
switching intention can be attributed to other factors not captured in the research model
or to random measurement error.
Table 3. Indirect Hypothesis Testing Results

Original Sample Standard T statistics P-values
sample (O) mean (M) deviation

EE -> TR -> SWI 0.069 0.069 0.016 4,195 0.000

PE -> TR -> SWI 0.061 0.061 0.016 3,782 0.000
Source: Research team survey results, 2025
The results show that Trust in AI significantly mediates the relationship between
Expectancy factors and Switching Intention. Specifically, the indirect effects of Effort
Expectancy and Performance Expectancy through Trust in AI are statistically significant (p
< 0.001), confirming the mediating role of trust.
3.2. Discussion
The results indicate that Push factors have no significant effect, implying that
dissatisfaction with traditional healthcare services is insufficient to trigger switching
behavior. Unlike low-risk services, where dissatisfaction often leads to change (Bansal et
al., 2005; Moon, 1995), decisions in healthcare are associated with high perceived risk,
habitual behavior, and trust-based relationships, leading individuals to prioritize stability
(Fastoso et al., 2012). In contrast, pull factors exert the strongest positive effect,
confirming the roles of performance expectancy, effort expectancy, and personal
innovativeness, consistent with UTAUT and TAM (Venkatesh et al., 2003) as well as prior
studies on AI in healthcare (Zhou, 2015; Agarwal & Prasad, 1998). Mooring factors
negatively influence switching intention through perceived risk, social influence, and
switching costs (Chen & Keng, 2019). Trust in AI has a positive effect and serves as a
mediating factor by reducing perceived risk, but it does not independently generate
switching motivation.
4. Conclusion and recommendations
This study makes two key contributions to the extension of the PPM framework in
smart healthcare. First, the PPM model operates asymmetrically, with switching intention
mainly driven by pull and mooring factors, while push effects are insignificant; users are
not leaving traditional healthcare due to dissatisfaction but are attracted by AI’s benefits,
though still constrained by psychological and social barriers. Second, trust in AI serves as a
context-dependent mediator, translating performance and effort expectations into
behavioural intention by reducing perceived risks, reflecting the healthcare context in
which adoption depends on proven clinical effectiveness, safety, and reliability.
Based on the above findings, this study proposes the following policy and practical
solutions. It is essential to strengthen legal frameworks and safety standards, invest in
digital infrastructure, and promote evidence-based communication; hospitals should
demonstrate AI’s clinical effectiveness and maintain physicians’ central role; healthcare

430