Development and Validation of an Automated Pediatric Cancer Staging Calculator Using the Toronto Pediatric Cancer Stage Guidelines
Bibliographic record
Abstract
Background: Pediatric cancer stage at diagnosis is critical for prognosis and research comparisons. The Toronto Pediatric Cancer Stage Guidelines standardize staging across childhood malignancies . We developed a framework for automated staging of pediatric cancers. Methods: A structured staging schema was created. An extraction pipeline was implemented to orchestrate agents. The system ingests multi-disciplinary meeting notes, pathology reports, radiology findings, operative notes, and clinic documentation from the first 3 months after diagnosis. One agent identifies the cancer type and maps it to a Toronto diagnostic category, after which another agent applies the relevant staging logic; and a validation agent examines the stage and its logic against summarized documentations. We tested the tool on 500 pediatric cancer cases from our institutional registry. Cases outside the Toronto schema (e.g. acute myeloid leukemia and nasopharyngeal carcinoma, which have no stage per guidelines) were excluded, yielding 433 evaluable cases. Each case was processed independently in two runs. The outputs were compared to an expert consensus reference stage (ground truth) established by four pediatric oncologists. Results: The automated system matched the reference stage in 91.2% of cases overall. Per-run accuracy (compared to ground truth) was 93.8% for the first run and 88.7% for the second run. The two runs agreed on 89.8% of cases (Cohen’s κ=0.785, p<0.001). Accuracy dropped significantly when the validation agent flagged the stage and requested a recalculation. For stages obtained from the first attempt, accuracy was 97%; while for stages achieved on subsequent attempts, accuracy achieved 77%. Conclusion: We demonstrate the first automated staging system for pediatric cancers using standardized Toronto criteria. The tool showed high accuracy comparable to human experts and excellent consistency between independent runs. We identified a measurable metric (number of calculation attempts) that can flag problematic cases for further human analysis.
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 | 0.010 | 0.027 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.004 | 0.002 |
| Science and technology studies | 0.001 | 0.000 |
| Scholarly communication | 0.002 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.003 | 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".