Education Research: Creating Online Interactive Case-Based Learning Experiences From Educational Case Reports With Large Language Models
Bibliographic record
Abstract
Background and Objectives: into an interactive online format to facilitate case-based learning. Methods: were converted into a free-text "screenplay" using the LLM Claude 3.5 Sonnet. These "screenplays" were then delivered in an interactive format through an online platform using GPT-4o. Two neurology fellows interrogated (prompted) the cases delivered by the online platform in a question-and-answer manner, seeking history, examination findings, and investigation results to arrive at a diagnosis and plan. These neurology fellows were not aware of the case report or screenplay content and asked questions in a manner that they would when evaluating a patient. A neurologist then reviewed each question-and-answer exchange for "screenplay" adherence and medical appropriateness. Results were analyzed with descriptive statistics. Results: The overall number of appropriate responses generated by the LLM was 206 of 210 (98.1%). There were 26 of 210 responses in which additional content was generated, all of which were medically plausible or consistent with the context of the case. The 4 errors that occurred were omissions of investigation results at the "screenplay" stage, which are amenable to manual correction. The omissions were the results of 3 unrevealing blood tests and 1 electroencephalogram result. None of these errors precluded the establishment of the diagnosis and completion of the case. Discussion: into an interactive question-and-answer format using LLMs. It should be noted that the nondeterministic nature of frontier LLMs and the potential for such LLM versions to change frequently are relevant considerations in making estimates of performance. Further studies investigating the impacts of this educational innovation are required.
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 | 0.023 | 0.076 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.002 | 0.003 |
| Scholarly communication | 0.006 | 0.008 |
| Open science | 0.004 | 0.008 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.022 | 0.005 |
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".