Breast Cancer Clinical NER Dataset

Dataset Description

This dataset contains annotated clinical texts in Spanish for Named Entity Recognition (NER) tasks specifically focused on breast cancer medical records. The dataset is designed to identify and classify medical entities relevant to breast cancer diagnosis, treatment, and patient management.

Dataset Summary

Language: Spanish
Task: Named Entity Recognition (NER)
Domain: Clinical/Medical - Breast Cancer
Total samples: 7,148 sentences (6,104 train + 522 validation + 522 test)
Dataset size: 1.64 MB
Download size: 389 KB
Format: BIO tagging scheme with 49 labels

Supported Tasks

Named Entity Recognition for clinical breast cancer texts
Medical information extraction
Clinical text processing and analysis

Dataset Structure

Data Fields

sentencia: A sequence of tokens (words) from clinical texts
tag: A sequence of BIO tags corresponding to each token

Data Splits

The dataset is split into three subsets:

Split	Number of examples	Size (bytes)
Train	6,104	1,357,988
Validation	522	142,275
Test	522	140,091
Total	7,148	1,640,354

Entity Types

The dataset includes 24 different entity types (48 BIO labels + 1 O label = 49 total labels) related to breast cancer clinical information:

Patient Demographics & History

AGE: Patient age information
FAMILY: Family history details
GINECOLOGICAL_HISTORY: Gynecological background
GINE_OBSTETRICS: Obstetric history
ALLERGIES: Known allergies

Clinical Assessment

STAGE: Cancer staging information
TNM: TNM classification system
BIOMARKER: Biomarkers and molecular markers
COMORBIDITY: Comorbid conditions
PROGRESION: Disease progression status

Temporal Information

DATE: Explicit dates
IMPLICIT_DATE: Implicit temporal references
DURATION: Duration of treatments or conditions

Treatment Information

TREATMENT_NAME: Names of treatments
LINE_CICLE_NUMBER: Treatment line and cycle numbers
SURGERY: Surgical procedures
DRUG: Medications and drugs
DOSE: Medication dosages
FREQ: Frequency of treatments

Lifestyle & Habits

TOXIC_HABITS: Smoking, alcohol, etc.
HABIT-QUANTITY: Quantities related to habits

Healthcare System

CLINICAL_SERVICE: Clinical departments/services
OCURRENCE_EVENT: Clinical events and occurrences

Cancer-Specific

CANCER_CONCEPT: General cancer-related concepts

Tagging Scheme

The dataset uses the BIO (Beginning-Inside-Outside) tagging scheme:

B-{ENTITY}: Beginning of an entity
I-{ENTITY}: Inside/continuation of an entity
O: Outside any entity (not part of a named entity)

Usage Example

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("anvorja/breast_cancer_es_dataset")

# Access different splits
train_data = dataset['train']        # 6,104 examples
validation_data = dataset['validation']  # 522 examples  
test_data = dataset['test']          # 522 examples

# View a sample
print(train_data[0])
# Output: {'sentencia': ['word1', 'word2', ...], 'tag': [0, 1, ...]}

# Get label names
labels = train_data.features['tag'].feature.names
print(f"Total labels: {len(labels)}")  # 49 labels
print("Sample labels:", labels[:5])

# Dataset statistics
print(f"Train size: {len(train_data)} examples")
print(f"Validation size: {len(validation_data)} examples") 
print(f"Test size: {len(test_data)} examples")

Dataset Creation

Source Data

The dataset appears to be created from Spanish clinical texts related to breast cancer cases, likely from medical records or clinical reports.

Annotation Process

The texts have been manually annotated using the BIO tagging scheme to identify relevant medical entities. The annotation follows clinical guidelines for breast cancer information extraction.

Considerations for Use

Medical Domain Expertise

This dataset requires domain knowledge in:

Clinical oncology
Spanish medical terminology
Breast cancer treatment protocols

Privacy and Ethics

Ensure compliance with medical data privacy regulations
Consider patient confidentiality when using this dataset
Follow institutional review board guidelines if applicable

Limitations

Dataset is specific to Spanish language clinical texts
Focused on breast cancer domain only
May not generalize to other cancer types or medical conditions

Citation

If you use this dataset in your research, please cite:

@dataset{breast_cancer_ner_es,
  title={Breast Cancer Clinical NER Dataset - Spanish},
  author={Borja, Carlos Andrés and Melo, Deisy Catalina},
  year={2024},
  publisher={Hugging Face},
  url={https://huggingface.co/datasets/anvorja/breast_cancer_es_dataset}
}

License

CC-BY-SA-4.0

Contact

For questions or issues regarding this dataset, please contact:

Carlos Andrés Borja
Deisy Catalina Melo

This dataset is intended for research and educational purposes in the field of clinical natural language processing and medical informatics.

Datasets:

anvorja
/

breast_cancer_es_dataset

You need to agree to share your contact information to access this dataset