French Oncology Pathology Reports With Genomic Biomarker Annotations

PARHAF-biomarkers-annotated is a French-language corpus of fictional clinical reports for oncology, created to support clinical NLP system development. This dataset contains training data for extracting genomic and tissue biomarkers from unstructured pathology reports. It was authored by HealthDataHub.

Use Cases

• Train a named entity recognition model to extract genomic biomarker mentions from French pathology report text.
• Develop a relation extraction system to characterize relationships between tissue biomarkers and clinical findings in oncology reports.
• Benchmark French clinical NLP systems on the specific task of biomarker identification from unstructured narratives.
• Fine-tune a language model for the oncology domain using annotated French clinical text focused on biomarkers.

Strengths

• Dataset is part of the larger PARHAF corpus, an open French corpus of human-authored clinical reports.
• Specifically designed for the development and evaluation of clinical NLP systems in oncology.
• Annotations target the extraction and characterization of genomic and tissue biomarkers, a key clinical task.

Limitations

• Dataset size, row count, column structure, and file formats are unknown from the provided input.
• Contains only training data; the test set is described as separate and not included here.
• Data consists of reports for fictional patients, which may not fully capture the complexity of real-world clinical language.

Provenance

Source

HealthDataHub, as a subpart of the PARHAF corpus.

Collection Method

Human-authored clinical reports of fictional patients, annotated for biomarker extraction.

Time Range

null

Freshness

Last updated on 2026-03-24.

Geography

France, based on language and tags.