README.md

---
license: mit
language:
- en
tags:
- clinical_trial
size_categories:
- 1M<n<10M
---

# TrialAlign
<p align="center">
  🤗 <a href="https://huggingface.co/linjc16/Panacea-7B-Chat" target="_blank">[Panacea Model] </a> • 💻 <a href="https://github.com/linjc16/Panacea" target="_blank">[Github Repo]</a> • 📃 <a href="https://arxiv.org/pdf/2407.11007" target="_blank">[Panacea Paper]</a> 
</p>

TrialAlign datasets are the continual pretraining corpus for "Panacea: A foundation model for clinical trial search, summarization, design, and recruitment". TrialAlign contains 793,279 clinical trial design documents worldwide and 1,113,207 clinical study papers.

## Load the dataset
```bash
pip install -U datasets fsspec
```

```python
from datasets import load_dataset
ds = load_dataset('linjc16/ctgov', split='train')
```

## Statistics of TrialAlign

| Source                          | Counts   |
|---------------------------------|----------|
| ClinicalTrials.gov              | 432,676  |
| ChiCTR (China)                  | 65,181   |
| EUCTR (EU)                      | 43,315   |
| JRCT (Japan)                    | 60,645   |
| ANZCTR (Australia/New Zealand)  | 23,374   |
| ISRCTN.org                      | 22,966   |
| ReBEC (Brazil)                  | 5,889    |
| CRIS (Korea)                    | 8,428    |
| DRKS (Germany)                  | 13,789   |
| IRCT (Iran)                     | 34,097   |
| TCTR (Thailand)                 | 7,443    |
| LTR (Netherlands)               | 9,768    |
| PACTR (Africa)                  | 3,848    |
| SLCTR (Sri Lanka)               | 421      |
| Trial Papers (Embase + PubMed)  | 1,113,207|



## Citation

If you find our work useful, please consider citing Panacea:

```
@article{lin2024panacea,
  title={Panacea: A foundation model for clinical trial search, summarization, design, and recruitment},
  author={Lin, Jiacheng and Xu, Hanwen and Wang, Zifeng and Wang, Sheng and Sun, Jimeng},
  journal={arXiv preprint arXiv:2407.11007},
  year={2024}
}
```