JMIR Publications

ABSTRACT

Background:

Case-based learning (CBL) promotes transfer of knowledge to practice, yet occupational health CBL must also develop exposure assessment and epidemiologic thinking. Static, single-session cases that disclose all information upfront can truncate iterative reasoning and encourage premature diagnostic closure. Generative AI (GenAI) can create scalable, high-fidelity scenarios and distractors, but hallucination risks require strict quality control.

Objective:

To develop and evaluate a GenAI-assisted, multi-stage scripted four-act CBL model with progressive disclosure for an occupational lead poisoning module, using a human-in-the-loop (HITL) workflow to mitigate hallucinations

Methods:

In a quasi-experimental pretest–posttest controlled study, 224 undergraduates were assigned by class to an intervention group (n=114) or control group (n=110). The control group received conventional static CBL; the intervention group received a four-act case with staged release of key information and instructor-verified distractor cues. GenAI materials were produced in a faculty-controlled environment and validated through iterative prompting, blinded expert review, and hallucination filtering/correction. Primary outcomes were standardized case-analysis assignment scores and delayed final-exam scores. Secondary outcomes were 0–10 ratings of learning experience; process outcomes from recordings included voluntary responses, discussion time, and evidence-referencing statements.

Results:

Baseline characteristics were comparable. The intervention group scored higher on case analysis (88.41±3.85 vs 82.51±3.94; P<0.001, d =1.51) and on the final exam (80.96±8.49 vs 78.62±6.63; P =0.0229, d =0.30). Significant improvements were observed in information gathering, hypothesis generation, and differential diagnosis (all P<0.001), while treatment/management planning did not differ (P=0.180). The intervention group reported higher perceived reasoning improvement, engagement, and transfer self-efficacy, alongside higher perceived difficulty (all P<0.001). Voluntary responses and discussion time increased; evidence referencing showed a nonsignificant upward trend. Audit logs indicated expert correction remained necessary.

Conclusions:

A GenAI-assisted, four-act progressive-disclosure CBL model with rigorous HITL verification can enhance exploratory clinical reasoning and engagement in occupational health education.