HiTZ/casimedicos-arg

---
license: cc-by-4.0
language:
- en
- es
- fr
- it
tags:
- casimedicos
- explainability
- medical exams
- medical question answering
- multilinguality
- argument mining
- argument generation
- LLMs
- LLM
pretty_name: CasiMedicos-Arg
configs:
- config_name: en
data_files:
- split: train
path:
- en/train_en_all.csv
- split: validation
path:
- en/validation_en_all.csv
- split: test
path:
- en/test_en_all.csv
- config_name: es
data_files:
- split: train
path:
- es/train_es_ordered.csv
- split: validation
path:
- es/validation_es_ordered.csv
- split: test
path:
- es/test_es_ordered.csv
- config_name: fr
data_files:
- split: train
path:
- fr/train_fr_ordered.csv
- split: validation
path:
- fr/validation_fr_ordered.csv
- split: test
path:
- fr/test_fr_ordered.csv
- config_name: it
data_files:
- split: train
path:
- it/train_it_ordered.csv
- split: validation
path:
- it/validation_it_ordered.csv
- split: test
path:
- it/test_it_ordered.csv
task_categories:
- text-generation
- question-answering
- token-classification
size_categories:
- 1K<n<10K
---

<p align="center">
<br>
<img src="http://www.ixa.eus/sites/default/files/anitdote.png" style="height: 200px;">
<br>

# CasiMedicos-Arg: A Medical Question Answering Dataset Annotated with Explanatory Argumentative Structures

[CasiMedicos-Arg](https://huggingface.co/datasets/HiTZ/casimedicos-arg) is, to the best of our knowledge, the first
multilingual dataset for Medical Question Answering where correct and incorrect diagnoses for a clinical case are
enriched with a natural language explanation written by doctors.
The [casimedicos-exp](https://huggingface.co/datasets/HiTZ/casimedicos-exp) have been manually annotated with
argument components (i.e., premise, claim) and argument relations (i.e., attack, support).
Thus, Multilingual CasiMedicos-arg dataset consists of 558 clinical cases (English, Spanish, French, Italian) with explanations,
where we annotated 5021 claims, 2313 premises, 2431 support relations, and 1106 attack relations.

<table style="width:33%">
<tr>
<th>Antidote CasiMedicos-Arg splits</th>
<tr>
<td>train</td>
<td>434</td>
</tr>
<tr>
<td>validation</td>
<td>63</td>
</tr>
<tr>
<td>test</td>
<td>125</td>
</tr>
</table>

- 📖 Paper:[CasiMedicos-Arg: A Medical Question Answering Dataset Annotated with Explanatory Argumentative Structures](https://aclanthology.org/2024.emnlp-main.1026/)
- 💻 Github Repo (Data and Code): [https://github.com/ixa-ehu/antidote-casimedicos](https://github.com/ixa-ehu/antidote-casimedicos)
- 🌐 Project Website: [https://univ-cotedazur.eu/antidote](https://univ-cotedazur.eu/antidote)
- Funding: CHIST-ERA XAI 2019 call. Antidote (PCI2020-120717-2) funded by MCIN/AEI /10.13039/501100011033 and by European Union NextGenerationEU/PRTR

## Example of Document in Antidote CasiMedicos Dataset

<p align="center">
<img src="https://github.com/ixa-ehu/antidote-casimedicos/blob/main/casimedicos-arg-example.png?raw=true" style="height: 600px;">
</p>

## Structure of CasiMedicos Dataset

```
id: question id,
text: tokenized text for the argument component detection (list of lists)
labels: labels for the argument component detection (list of lists)
relations: relations between arguments (list of <argument1, argument2, relation> triplets)
```

## Results of Argument Component Detection using LLMs

<p align="left">
<img src="https://github.com/ixa-ehu/antidote-casimedicos/blob/main/multingual-data-transfer.png?raw=true" style="height: 400px;">
</p>

## Citation

If you use CasiMedicos-Arg then please **cite the following paper**:

```bibtex
@inproceedings{sviridova-etal-2024-casimedicos,
title = {{CasiMedicos-Arg: A Medical Question Answering Dataset Annotated with Explanatory Argumentative Structures}},
author = "Sviridova, Ekaterina and
Yeginbergen, Anar and
Estarrona, Ainara and
Cabrio, Elena and
Villata, Serena and
Agerri, Rodrigo",
booktitle = "Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing",
year = "2024",
url = "https://aclanthology.org/2024.emnlp-main.1026",
pages = "18463--18475"
}
```

**Contact**: [Rodrigo Agerri](https://ragerri.github.io/)
HiTZ Center - Ixa, University of the Basque Country UPV/EHU