The dataset is currently empty. Upload or create new data files. Then, you will be able to explore them in the Dataset Viewer.
YAML Metadata Warning:empty or missing yaml metadata in repo card
Check out the documentation for more information.
MedSynth: Clinical Note Synthesis
π Paper | π Project Page | πΎ Released Resources | π¦ Repo
This is the resource page of our clinical note synthesis collection on Huggingface.
Dataset
| Dataset | Link |
|---|---|
| MedSynth-ClinicalNotes | π€ |
Please also check the raw data after our processing if you are interested: toolevalxm/MedSynth-ClinicalNotes-Raw.
Models
| Base Model / Training | MedSynth | MedSynth++ | ||
|---|---|---|---|---|
| Stage 1 | Stage 2 | Stage 1 | Stage 2 | |
| Mistral 7B Medical | π€ | π€ | π€ | π€ |
| LLaMA 3.1 8B | π€ | π€ | π€ | π€ |
| Meditron 7B | π€ | π€ | π€ | π€ |
Introduction
While clinical notes contain rich medical information, generating realistic synthetic clinical data presents unique challenges: 1) Maintaining clinical accuracy while ensuring patient privacy; 2) Generating coherent narratives that reflect real clinical workflows. Thus, we adopt a fully LLM-based approach for synthesizing all clinical notes using GPT-4o, leveraging its superior medical knowledge and instruction-following capabilities.
*Due to healthcare privacy requirements, we only release the fully synthetic subset (this page) of the full MedSynth dataset.
License
The license for this dataset is CC0-1.0.
- Downloads last month
- 5