EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Automatic extraction of clinically relevant parameters for tumor board decision support

Y&#xFC, z&#xFC, nc&#xFC, oglu, Imge, Chen, Yuxuan, L&#xFC, ser, A. Altar, Ramasetti, Nikitha Shruthi und 11 weitere

GMS Medizinische Informatik, Biometrie und Epidemiologie · 22.09.2026 · 22 (Doc13) · Doc13

DOI: 10.3205/mibe000311 ↗

Kurz gefasst

Fachartikel in der Zeitschrift GMS Medizinische Informatik, Biometrie und Epidemiologie, erschienen am 22.09.2026. Verfasst von Y&#xFC und 22 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Introduction

Clinical decision-making relies on information documented in unstructured reports, which are often non-standardized and difficult to process efficiently. This challenge is particularly critical in multidisciplinary tumor board meetings, where decisions must be made under time constraints and based on incomplete information. Automated information extraction (IE) as presented in ADBoard supports structuring relevant clinical data for decision-making.

Methods

We developed and evaluated an IE pipeline for liver tumor cases, focusing on hepatocellular carcinoma, cholangiocarcinoma, and colorectal liver metastases. A dataset of 743 clinical documents from 134 patients was manually annotated to create ground truth labels for clinically relevant parameters. The task was formulated as named entity recognition (NER). We compared rule-based methods (regular expressions), transformer-based models, and an exploratory large language model (LLM) under realistic clinical constraints.

Results

Transformer-based models achieved the strongest overall performance, particularly for frequent parameters. Regular expressions performed competitively for well-defined patterns and proved robust in low-resource settings. The LLM showed inconsistent performance and did not outperform the other methods. Across all approaches, performance was also affected by data imbalance and annotation variability.

Discussion

Our findings indicate that, under the evaluated realistic, local deployment constraints, the exploratory LLaMA 3.1 8B setup did not outperform task-specific transformer or rule-based approaches for specialized clinical IE. Instead, a hybrid approach combining transformer-based and rule-based methods might be most effective. Furthermore, evaluation methodology substantially influences performance interpretation, as strict span-based metrics can underestimate clinically relevant results. These results highlight that improving annotation quality and evaluation design is as important as advances in model architecture for clinical IE.

Schlagwörter

natural language processinginformation extractionhepatocellular carcinomacholangiocarcinomacolorectal liver metastasessequence evaluationtumor boardclinical decision supportnatürliche SprachverarbeitungInformationsextraktion

Zitieren

APA

Y&#xFC, z&#xFC, nc&#xFC, oglu, Imge, Chen, Yuxuan, L&#xFC, ser, Altar, A., Ramasetti, Shruthi, N., Oehring, Robert, Krenzien, Felix, Thomas, Philippe, M&#xF6, …, & Roland (2026). Automatic extraction of clinically relevant parameters for tumor board decision support. GMS Medizinische Informatik, Biometrie und Epidemiologie, 22(Doc13), Doc13. https://doi.org/10.3205/mibe000311

Vancouver

Y&#xFC, z&#xFC, nc&#xFC, oglu, Imge, Chen, et al. Automatic extraction of clinically relevant parameters for tumor board decision support. GMS Medizinische Informatik, Biometrie und Epidemiologie. 2026;22(Doc13):Doc13. doi:10.3205/mibe000311

BibTeX
@article{YxFC2026,
  author = {Y&#xFC and z&#xFC and nc&#xFC and oglu and Imge and Chen and Yuxuan and L&#xFC and ser and Altar, A. and Ramasetti and Shruthi, Nikitha and Oehring and Robert and Krenzien and Felix and Thomas and Philippe and M&#xF6 and ller and Sebastian and Roller and Roland},
  title = {Automatic extraction of clinically relevant parameters for tumor board decision support},
  journal = {GMS Medizinische Informatik, Biometrie und Epidemiologie},
  year = {2026},
  volume = {22},
  number = {Doc13},
  pages = {Doc13},
  doi = {10.3205/mibe000311}
}
Kostenlos starten →