Evaluation of a Problem-Based Learning Program’s Effect on Artificial Intelligence Ethics Among Japanese Medical Students: Mixed Methods Study
Bibliographic record
Abstract
BACKGROUND: The rapid advancement of artificial intelligence (AI) has had a substantial impact on medicine, necessitating the integration of AI education into medical school curricula. However, such integration remains limited. A key challenge is the discrepancy between medical students' positive perceptions of AI and their actual competencies, with research in Japan identifying specific gaps in the students' competencies in understanding regulations and discussing ethical issues. OBJECTIVE: This study evaluates the effectiveness of an educational program designed to improve medical students' competencies in understanding legal and ethical AI-related issues. It addresses the following research questions: (1) Does this educational program improve students' knowledge of AI and its legal and ethical issues, and what is each program element's contribution to this knowledge? (2) How does this educational program qualitatively change medical students' thoughts on these issues from an abstract understanding to a concrete and structured thought process? METHODS: This mixed methods study used a single-group pretest and posttest framework involving 118 fourth-year medical students. The 1-day intervention comprised a lecture and problem-based learning (PBL) session centered on a clinical case. A 24-item multiple-choice questionnaire (MCQ) was administered at 3 time points (pretest, midtest, and posttest), and descriptive essays were collected before and after the intervention. Data were analyzed using linear mixed-effects models, the Wilcoxon signed-rank test, and text mining, including comparative frequency analysis and cooccurrence network analysis with Jaccard coefficients. An optional survey on student perceptions based on the attention, relevance, confidence, and satisfaction model was conducted (n=76, 64.4%). RESULTS: Objective knowledge scores increased significantly from the pretest (median 17, IQR 15-18) to posttest (median 19, IQR 17-21; β=1.42; P<.001). No significant difference was observed between score gains during the lecture and PBL phases (P=.54). Qualitative text analysis revealed the significant transformation of cooccurrence network structures (Jaccard coefficients 0.116 and 0.121) from fragmented clusters to integrated networks. Students also used professional and ethical terminology more frequently. For instance, use of the term "bias" in patient explanations increased from 10 (8.5%) at pretest to 25 (21.2%) at posttest, while references to "personal information" in physician precautions increased from 36 (30.5%) to 50 (42.4%). The optional survey indicated that students' confidence (mean 3.78, SD 0.87) was significantly lower than their perception of the program's relevance (mean 4.20, SD 0.71; P<.001). CONCLUSIONS: This PBL-based program was associated with the improvements in knowledge and, more importantly, a structural transformation in students' thinking about AI ethics from an abstract level to a concrete, clinically grounded reasoning. The discrepancy between quantitative and qualitative findings suggests limitations of MCQs in assessing higher-order skills fostered by PBL. Overall, this study indicates the potential of PBL as an effective pedagogical method for AI ethics education.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.012 | 0.014 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".