AI-Generated Microlearning for Plastic Surgery Residency: Single-Arm Pre-Post Feasibility Study.

Abstract

Background: Surgical residency training faces mounting pressure from expanding subspecialty knowledge requirements alongside compressed learning opportunities. Microlearning, defined as brief, focused educational episodes, offers a pedagogically sound approach suited to fragmented clinical schedules, while large language models present capabilities for scalable content generation. However, the feasibility, quality, and educational value of large language model-generated microlearning in surgical training remain unexplored.

Objective: This study aimed to assess the feasibility and acceptability of artificial intelligence (AI)-generated microlearning modules among plastic surgery residents, evaluate content quality through systematic faculty evaluation, and generate preliminary effect size estimates for self-perceived knowledge and confidence changes.

Methods: A single-arm pilot feasibility study with a pre-post design was conducted at an academic plastic surgery residency program (N=11 residents) over 12 weeks. Six modules covering core subspecialties were developed using Google Gemini 2.5 Pro. From 60 AI-generated multiple-choice questions, 42 were selected following independent evaluation by 2 board-certified plastic surgeons using a 5-dimension rubric (accuracy, relevance, clarity, pedagogical value, and safety). Modules were released biweekly via Google Forms. Outcomes included retention, acceptability (7-point Likert scales), faculty evaluation ratings, item-level psychometrics, and pre-post changes in self-perceived knowledge and confidence. Composite scores were calculated as the mean change across all 6 subspecialty topics. Wilcoxon signed-rank tests served as the primary analysis for matched pairs (n=9). Analyses were conducted using Python 3.11.

Results: All 11 residents completed postassessments. Acceptability was high across all 6 subitems (all medians ≥5 out of 7). Faculty evaluation confirmed high content quality (mean quality index 4.68/5); interrater reliability was poor (intraclass correlation coefficient (3,1)=0.231). Among 9 matched pairs, composite confidence improved significantly (median change 0.33; d=0.65; P=.047), while composite knowledge showed a medium effect (d=0.71; P=.09). Topic-specific gains were largest where baseline was lowest: hand surgery knowledge (d=1.48; P=.02) and confidence (d=1.41; P=.02). Resource use and sense-of-community measures remained stable.

Conclusions: AI-generated microlearning modules are feasible and acceptable for plastic surgery residency education. Faculty-evaluated content achieved high quality ratings (mean 4.68/5), and residents showed strong engagement with preliminary improvements in self-perceived confidence. The intervention was integrated into existing learning ecosystems without disrupting traditional resources or peer dynamics. These findings provide effect size estimates to support the design of adequately powered randomized trials.