Performance of ChatGPT-4o in Determining Radiology–Pathology Concordance and Management Recommendations Following Image-Guided Breast Biopsies
Bibliographic record
Abstract
Background: Determining radiology–pathology concordance after breast biopsies is critical to ensuring appropriate patient management. However, expertise and multidisciplinary input are not universally accessible. Purpose: To evaluate the performance of a large language model, ChatGPT-4o, in determining the radiology–pathology concordance of breast biopsies and suggesting subsequent management steps. Methods: A retrospective single-center study analyzed 244 cases of image-guided breast biopsies of women. ChatGPT-4o assessed de-identified radiology and pathology reports for concordance and recommended management. Radiologist assessments served as the reference standard with final surgical pathology and 2-year imaging follow-up serving as gold standards when applicable. Concordance rates, management recommendations, and diagnostic agreement with the gold standard were compared using statistical tests, including McNemar’s, chi-square, Fisher–Freeman–Halton, and Cohen’s kappa. Results: ChatGPT-4o achieved a concordance rate of 98.8% vs. 98.0% for radiologists (p = 0.625) and demonstrated high diagnostic agreement with the gold standard (kappa = 0.947, p < 0.001). ChatGPT-4o favored imaging follow-up more than radiologists (49.2% vs. 41.8%, p < 0.001) and surgical management less frequently (41.8% vs. 46.7%). Conclusions: ChatGPT-4o demonstrated diagnostic performance comparable to radiologists with breast imaging subspecialities in evaluating breast biopsy concordance. Its slightly more conservative management approach may enhance shared decision-making in resource-limited settings.
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 | 0.000 | 0.001 |
| 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".