158eP Clustering and response prediction in differentiated thyroid cancer: Insights from the Hungarian thyroid cancer register
Bibliographic record
Abstract
Background: Endometrial cancer (EC) accounted for 420,368 new cases globally in 2022, with China, the US, and Russia reporting the highest incidence.Rising case numbers and evolving multi-modality treatments underscore the need for innovative, less invasive, and resource-efficient diagnostic approaches.Radiogenomics-linking imaging features with molecular profiles-offers such potential.This study evaluated whether natural language processing (NLP) applied to MRI reports can predict EC molecular subtypes.Methods: Anonymised MRI reports from 16 EC patients were analysed using EndoMAPP, a lightweight ClinicalBERT-based NLP model built indigenously using the python coding language.This system parsed PDF reports and generated real-time predictions of molecular subtypes with interpretability.Results: For POLE mutation status, the model achieved precision of 78%, recall of 90%, and F1-score of 84%.Performance for p53 abnormalities and MMR deficiency was moderate but consistent, with precision and recall around 75-78% and aligned F1-scores. Conclusions:These results indicate balanced detection while identifying the need for further optimisation, particularly in distinguishing p53-abnormal EC, which is linked to high-risk disease and adverse outcomes.ClinicalBERT-based NLP applied to routine imaging reports shows promise as a radiogenomic tool for non-invasive molecular classification in EC.This approach enables rapid, scalable, and interpretable molecular prediction from standard-of-care documents, supporting precision oncology and early risk stratification in gynaecological malignancies especially in resource limited centres.A concordance study is planned for patients with molecular reports available prior to surgery, as well as for those who underwent molecular NGS profiling, in order to make the model more robust.Legal entity responsible for the study: D.K. Gudipudi.
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 | 0.001 | 0.008 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.006 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.000 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
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".