An AI-based online learning platform improves blood cell morphology skills in medical laboratory technologists: a sequential explanatory study

作者：Junxun Li, Zhenhua Gao, 杨大雅, Fan Zhang, Runzhao Li, Yang Wang, Juan Ouyang, Bin Huang, Peisong Chen, Wange Lv, Ming Kuang, David Taylor · 发表于：BMC Medical Education · 年份：2026 · DOI：10.1186/s12909-026-10188-9 · 研究领域：Clinical Laboratory Practices and Quality Control、Bacterial Identification and Susceptibility Testing、Artificial Intelligence in Healthcare and Education

Blood cell morphology training requires repetitive, feedback-rich practice, whereas conventional continuing professional development (CPD) relies largely on static materials. Rigorous evidence for theory-informed, AI-supported skill-based CPD remains limited. We evaluated whether the RuiZhiXue (RZX) platform, informed by Vygotsky’s zone of proximal development (ZPD) and operationalised through intelligent distractors and real-time AI-generated probability feedback, improves morphology skills among medical laboratory technologists compared with conventional textbook- and atlas-based self-directed learning. We used an explanatory sequential mixed-methods design. Fifty-two medical laboratory technologists were randomised to RZX ( n = 26) or conventional textbook- and atlas-based self-directed learning ( n = 26) for two weeks. Parallel pretest and posttest assessments, each consisting of 50 image-based MCQs drawn from a validated test bank strictly separated from practice items, were auto-scored. Semi-structured interviews with 16 technologists and three mentors were subsequently analysed thematically to help interpret the quantitative findings. Both groups improved from pretest to posttest, but the RZX group achieved a significantly higher mean posttest score than the conventional self-directed learning group (86.62 vs. 64.77, p < 0.001), with a very large observed between-group effect size (Cohen’s d = 2.16). The RZX group also showed lower posttest-score variance ( p = 0.01), ...