Development and Validation of an Automated Pediatric Cancer Staging Calculator Using the Toronto Pediatric Cancer Stage Guidelines
Pediatric Blood & Cancer · 2026-05 · 73 (5)
DOI: 10.1002/pbc.32086 ↗Kurz gefasst
Validierungsstudie in der Zeitschrift Pediatric Blood & Cancer, online erschienen am 15.03.2026. Verfasst von Iyad Sultan und 8 weiteren Autor:innen. Beteiligte Einrichtung: Artificial Intelligence Office, King Hussein Cancer Center, Amman, Jordan. Hauptthemen laut MEDLINE: Neoplasm Staging, Neoplasms, Large Language Models, Natural Language Processing. Die Arbeit nennt 11 Literaturangaben.
Abstract (Originalsprache)
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 using electronic health records.
Methods
An extraction pipeline was implemented to orchestrate agents. The system ingests notes from the first 3 months after diagnosis. Agents select the most appropriate staging schema based on diagnosis, calculate the stage, and then validate its accuracy. We tested the tool on 500 pediatric patients from our institutional registry. Cases outside the Toronto schema 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 failed the first attempt and requested a recalculation. For stages obtained from the first attempt, accuracy was 97%; while for stages achieved on subsequent attempts, accuracy was 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 propose a hybrid workflow that flags uncertain cases for human review, yielding approximately 97% accuracy with approximately 30% human adjudication.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Stichwörter der Autor:innen
Autor:innen und Institutionen
- Artificial Intelligence Office, King Hussein Cancer Center, Amman, Jordan.
- Department of Pediatrics, King Hussein Cancer Center, Amman, Jordan.
- Faculty of Medicine, University of Jordan, Amman, Jordan.
- Department of Medicine, Jefferson Einstein Philadelphia Hospital, Philadelphia, Pennsylvania, USA.
- Office For Research and Scientific Affairs (OSAR), King Hussein Cancer Center, Amman, Jordan.
- Director General Office, King Hussein Cancer Center, Amman, Jordan.
ORCID-Profile: Iyad Sultan, Anwar Al-Nassan
Zitieren
APA
Sultan, I., Al-Nassan, A., Alomari, L., Altalla, B., Bazzeh, F., Halalsheh, H., Zandaki, D., Al-Omari, A., & Mansour, A. H. (2026). Development and Validation of an Automated Pediatric Cancer Staging Calculator Using the Toronto Pediatric Cancer Stage Guidelines. Pediatric Blood & Cancer, 73(5). https://doi.org/10.1002/pbc.32086
Vancouver
Sultan I, Al-Nassan A, Alomari L, Altalla B, Bazzeh F, Halalsheh H, et al. Development and Validation of an Automated Pediatric Cancer Staging Calculator Using the Toronto Pediatric Cancer Stage Guidelines. Pediatric Blood & Cancer. 2026;73(5). doi:10.1002/pbc.32086
BibTeX
@article{Sultan2026,
author = {Sultan, Iyad and Al-Nassan, Anwar and Alomari, Laith and Altalla, Bayan and Bazzeh, Faiha and Halalsheh, Hadeel and Zandaki, Dua'a and Al-Omari, Amal and Mansour, Asem H},
title = {Development and Validation of an Automated Pediatric Cancer Staging Calculator Using the Toronto Pediatric Cancer Stage Guidelines},
journal = {Pediatric Blood & Cancer},
year = {2026},
volume = {73},
number = {5},
doi = {10.1002/pbc.32086}
}
EUVIMED