Effectiveness of a Structured Artificial Intelligence Literacy Training Programme Among Community Health Nurses: A Simulated Two-Arm Controlled Experimental Study

Background: Evidence on how to build artificial intelligence (AI) literacy among community health nurses is scarce. Studies of nurses' AI literacy are mostly cross-sectional, and intervention studies that use multidimensional AI-literacy outcomes in community nursing settings appear to be lacking. Objective: To demonstrate, using a simulated dataset, the analytical workflow for evaluating a proposed structured AI-literacy training programme for community health nurses (change in AI literacy immediately after training and retention at 8 weeks), and to specify the design of a future prospective trial. Methods: This is a simulation study: no nurses were recruited and the programme has not been delivered. The proposed trial would be a two-arm, parallel-group, assessor-blinded randomised design with assessments at baseline (T0), immediately after training (T1) and at 8 weeks (T2). To illustrate the analysis, a dataset of 120 simulated nurses (60 per arm) was generated under assumed training effects and analysed with Welch's t-test on change scores, baseline-adjusted ANCOVA and a mixed-effects model. AI literacy was a 14-item, 7-domain score (14–70) adapted from the SAIL-HCW. Group differences are reported with 95% confidence intervals (CIs). Results: In the simulated dataset, mean AI literacy rose by 12.08 points (SD 4.57) in the intervention arm and by 1.35 points (SD 3.25) in the control arm. The between-group difference in change was 10.73 points (95% CI 9.29 to 12.17); the baseline-adjusted difference was 10.77 points (95% CI 9.31 to 12.23), and the unadjusted difference in 8-week scores was 12.53 points (95% CI 10.05 to 15.01). These estimates follow from the effects assumed when the data were generated and are not evidence of effectiveness. Conclusion: The simulation illustrates the analysis plan and reporting format for a future trial; it does not show that the programme improves AI literacy. Prospective, registered empirical testing with a pre-calculated sample size, objective competency assessment and longer follow-up is required.