Hugging Face's logo

Datasets:
gallifantjack
/
demo_onc_amie

Tasks:
Text Classification
Other
Modalities:
Tabular
Text
Formats:
parquet
Languages:
English
Size:
1K - 10K
Libraries:
Datasets
pandas
License:
Dataset card Viewer Files Community
1
demo_onc_amie / README.md
gallifantjack's picture
gallifantjack
Upload README.md with huggingface_hub
6aafeb1 verified
preview code
|
raw
history blame contribute delete
2.67 kB
metadata
language: en
license: mit
pretty_name: Demographic Variations Oncology AMIE Dataset
size_categories:
  - n<1K
task_categories:
  - text-classification
  - other
task_ids:
  - text-classification
  - medical-diagnosis

Demographic Variations Oncology AMIE Dataset

Dataset Description

This dataset extends the original Oncology AMIE dataset with demographic variations to enable robustness and bias evaluation. Each clinical case is replicated across different demographic combinations while preserving the medical content.

Dataset Summary

A clinical dataset containing breast cancer cases with demographic variations, including their descriptions, molecular phenotypes, and treatment recommendations. The dataset includes both treatment-naive and treatment-refractory cases.

Demographic Variations

The dataset includes systematic variations across:

  • Race: White, Black, Hispanic, Asian, American Indian
  • Income levels: High, Middle, Low
  • Geographic locations (ZIP codes):
    • 90210 (Beverly Hills - high income)
    • 10027 (Harlem, NY - mixed income)
    • 48207 (Detroit - lower income)
    • 98104 (Seattle - urban diverse)
    • 78577 (Rio Grande Valley - Hispanic majority)

Features

The dataset includes:

  • id: Unique identifier for the demographic variation version of the case
  • case_id: Original case identifier
  • variation: Demographic variation identifier
  • age: Patient age
  • sex: Patient sex
  • race: Patient race
  • income: Income level
  • zipcode: Geographic location
  • description: Detailed case description
  • type: Treatment-naive or refractory
  • molecular_phenotype: ER/PR/HER2 status
  • treatment_status: Current treatment status
  • special_considerations: Additional clinical considerations
  • ground_truth: Expert-provided treatment recommendation
  • system_prompt: AI system instruction
  • question_all: Complete case evaluation question
  • question_neoadjuvant: Specific question about neoadjuvant therapy
  • question_surgery: Specific question about surgical approach
  • question_adjuvant: Specific question about adjuvant therapy
  • question_genetic: Specific question about genetic testing
  • question_support: Specific question about psychosocial support

Usage

This dataset can be used for:

  • Evaluating model robustness across demographic variations
  • Assessing potential biases in medical AI systems
  • Clinical decision support development
  • Medical education and training
  • Research in oncology treatment protocols

Citation

If you use this dataset, please cite:

@misc{demo_onc_amie_dataset,
  title={Demographic Variations Oncology AMIE Dataset},
  author={Gallifant, Jack},
  year={2024},
  publisher={Hugging Face}
}