Related Experiment Video
Updated: Sep 24, 2026
Reducing State Anxiety Using Working Memory Maintenance
Published on: July 19, 2017
Daily temporal associations among AI replacement anxiety, state anxiety, and academic burnout indicators in medicine
Haoqian Li1,2, Zheng Zhang3, Bo Yang3,4
1Department of Health Management Center, Third Xiangya Hospital, Central South University, Changsha, 410011, China. 936557035@qq.com.
Background:
Artificial intelligence (AI) is increasingly entering medicine and nursing education, raising concerns about how healthcare students respond to AI-related occupational uncertainty. AI replacement anxiety (ARA) may become more or less salient across academic days, but little is known about its short-term within-person temporal associations with state anxiety and day-level academic burnout indicators.
Methods:
This 10-day weekday diary study included 751 medicine and nursing students who provided 4,625 observed diary records after quality screening. Daily surveys assessed ARA and three academic burnout constructs using single day-adapted proxy indicators, together with a two-item state-anxiety measure. Two-level Bayesian dynamic structural equation modeling tested six primary dimension-specific models in the ARA-to-state-anxiety and state-anxiety-to-ARA directions. Supplementary joint multivariate models, a stricter 695-participant sensitivity sample, and a targeted calendar-spacing sensitivity analysis examined unique effects, posterior contrasts, and robustness.
Results:
In the ARA-to-state-anxiety direction, higher daily ARA was associated with higher subsequent cynicism, and higher cynicism was associated with later state anxiety. This M2 component pattern remained supported when the weekend interval was represented using calendar-day spacing. In the reverse direction, higher state anxiety was consistently associated with a higher subsequent score on the academic-inefficacy proxy indicator, but the subsequent association of this indicator with ARA was less robust across model specifications and was attenuated in the weekend-spacing sensitivity analysis. The model-based indirect effects remained uncertain, with 95% credible intervals including zero. Joint models indicated that the clearest cross-dimension differences were concentrated mainly in first-stage associations rather than complete unique downstream indirect processes.
Conclusions:
The findings support dimension-specific temporal associations among ARA, state anxiety, and academic burnout-related daily indicators, highlighting cynicism and academic-inefficacy-related daily appraisals as particularly relevant components of the temporal links between ARA and state anxiety among medicine and nursing students. These findings provide a more differentiated basis for future research on targeted support and intervention in health-professions education.
