|
--- |
|
title: Clinical Report Generator |
|
emoji: 🏥 |
|
colorFrom: purple |
|
colorTo: blue |
|
sdk: gradio |
|
sdk_version: 5.13.2 |
|
app_file: app.py |
|
pinned: false |
|
license: mit |
|
short_description: Professional Clinical Report Generation using T5 |
|
--- |
|
|
|
# Clinical Report Generator |
|
|
|
This Hugging Face Space provides a clinical report generation service using a fine-tuned T5 model. The model has been trained on a dataset of clinical documents to generate professional, structured clinical reports from input notes. |
|
|
|
## Features |
|
|
|
- Generates professional clinical reports from input notes |
|
- Maintains medical terminology and context |
|
- Removes unwanted elements like URLs |
|
- Provides a user-friendly interface |
|
|
|
## How to Use |
|
|
|
1. Enter your clinical notes in the input text box |
|
2. Click "Submit" to generate the report |
|
3. The generated report will appear in the output text box |
|
|
|
## Model Details |
|
|
|
- Base Model: T5-small |
|
- Fine-tuned on: Clinical documents and structured templates |
|
- Input Length: Up to 512 tokens |
|
- Output Length: Up to 256 tokens |
|
- Special Features: |
|
- Beam search with 4 beams |
|
- No repeat ngram size of 3 |
|
- Length penalty of 2.0 |
|
- Early stopping enabled |
|
- URL filtering |
|
|
|
## Examples |
|
|
|
The interface includes example inputs to demonstrate the model's capabilities. These examples cover common clinical scenarios such as: |
|
- Initial patient presentations |
|
- Follow-up visits |
|
- Treatment responses |
|
|
|
## Notes |
|
|
|
- The model is designed for generating clinical reports only |
|
- Input should be clear and contain relevant clinical information |
|
- Generated reports should be reviewed by healthcare professionals |
|
|
