Datasets:

OpenMed
/

DDI-Corpus-Processed

license: cc-by-nc-4.0
task_categories:
  - text-classification
language:
  - en
tags:
  - medical
  - drug-drug-interaction
  - pharmacology
  - clinical
  - ddi
size_categories:
  - 1K<n<10K

DDI Corpus Processed

Drug-Drug Interaction (DDI) detection dataset processed from the SemEval-2013 DDI Corpus.

Dataset Description

This dataset contains drug-drug interaction examples for 5-class classification:

MECHANISM: Mechanistic description of how drugs interact (e.g., inhibition, induction)
EFFECT: Clinical effect of concurrent use (e.g., increased toxicity, decreased efficacy)
ADVISE: Advisory or cautionary information about concurrent use
INT: General interaction without specific type
NONE: No interaction between the drug pair

Dataset Statistics

Split	Total	MECHANISM	EFFECT	ADVISE	INT	NONE
Train	5,744	1,319 (23%)	1,687 (29%)	826 (14%)	189 (3%)	1,723 (30%)
Test	1,398	302 (22%)	360 (26%)	221 (16%)	96 (7%)	419 (30%)

Data Format

Each example contains:

sentence: The original sentence containing drug mentions
drug1: First drug entity
drug2: Second drug entity
relation: Interaction type (MECHANISM, EFFECT, ADVISE, INT, or NONE)
source_file: Original XML source file
sentence_id: Original sentence ID
pair_id: Original pair ID

Usage

from datasets import load_dataset

dataset = load_dataset("MaziyarPanahi/DDI-Corpus-Processed")

Citation

If you use this dataset, please cite:

@article{herrero2013ddi,
  title={The DDI corpus: An annotated corpus with pharmacological substances and drug--drug interactions},
  author={Herrero-Zazo, Mar{\'\i}a and Segura-Bedmar, Isabel and Mart{\'\i}nez, Paloma and Declerck, Thierry},
  journal={Journal of Biomedical Informatics},
  volume={46},
  number={5},
  pages={914--920},
  year={2013}
}

License

CC BY-NC 4.0 (following original DDI Corpus license)