Large Language Models in the Diagnosis of Hand and Peripheral Nerve Injuries: An Evaluation of ChatGPT and the Isabel Differential Diagnosis Generator
Bibliographic record
Abstract
Purpose: Tools using artificial intelligence may help reduce missed or delayed diagnoses and improve patient care in hand surgery. This study aimed to compare and evaluate the performance of two natural language processing programs, Isabel and ChatGPT-4, in diagnosing hand and peripheral nerve injuries from a set of clinical vignettes. Methods: Cases from a virtual library of hand surgery case reports with no history of trauma or previous surgery were included in this study. The clinical details (age, sex, symptoms, signs, and medical history) of 16 hand cases were entered into Isabel and ChatGPT-4 to generate top 10 differential diagnosis lists. Isabel and ChatGPT-4's inclusion and median rank of the correct diagnosis within each list were compared. Two hand surgeons were then provided each list and asked to independently evaluate the performance of the two systems. Results: Isabel correctly identified 7/16 (44%) cases with a median rank of two (interquartile range = 3). ChatGPT-4 correctly identified 14/16 (88%) of cases with a median rank of one (interquartile range = 1). Physicians one and two, respectively, preferred the lists generated by ChatGPT-4 in 12/16 (75%) and 13/16 (81%) of cases and had no preference in 2/16 (13%) cases. Conclusions: < .05) and generated higher quality differential diagnoses than Isabel. Isabel produced several inappropriate and imprecise differential diagnoses. Clinical relevance: Despite large language models' potential utility in generating medical diagnoses, physicians must continue to exercise high caution and use their clinical judgment when making diagnostic decisions.
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 teacher head, 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".