Spaces:
Sleeping
Sleeping
Update README.md
Browse files
README.md
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
---
|
| 2 |
title: Medical Document Summarizer
|
| 3 |
-
emoji:
|
| 4 |
-
colorFrom:
|
| 5 |
-
colorTo:
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 5.23.2
|
| 8 |
app_file: app.py
|
|
@@ -11,4 +11,35 @@ license: mit
|
|
| 11 |
short_description: Upload your files and get a brief summary!
|
| 12 |
---
|
| 13 |
|
| 14 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
---
|
| 2 |
title: Medical Document Summarizer
|
| 3 |
+
emoji: 🔥
|
| 4 |
+
colorFrom: yellow
|
| 5 |
+
colorTo: green
|
| 6 |
sdk: gradio
|
| 7 |
sdk_version: 5.23.2
|
| 8 |
app_file: app.py
|
|
|
|
| 11 |
short_description: Upload your files and get a brief summary!
|
| 12 |
---
|
| 13 |
|
| 14 |
+
# Medical Document Summarizer
|
| 15 |
+
|
| 16 |
+
This project is designed to automatically extract and summarize key information from clinical trial documents (e.g., PDF files of research articles) using state-of-the-art NLP models. The pipeline leverages the BigBird-Pegasus model for long-form summarization and includes content filtering, text cleaning, and post-processing to produce concise bullet-point and paragraph summaries.
|
| 17 |
+
|
| 18 |
+
## Features
|
| 19 |
+
|
| 20 |
+
*Note*: User has to upload medical document into the file directory before running the model.
|
| 21 |
+
|
| 22 |
+
- **PDF Extraction:** Reads and filters PDF files to capture only pages with core content (e.g., Abstract, Methods, Results, Conclusions).
|
| 23 |
+
- **Text Cleaning:** Removes noisy metadata, citations, and excess whitespace.
|
| 24 |
+
- **Core Section Extraction:** Attempts to identify and extract important sections using regex; falls back to header removal when sections are not detected.
|
| 25 |
+
- **Chunking & Summarization:** Splits the text into manageable chunks and uses the BigBird-Pegasus summarization model for each chunk.
|
| 26 |
+
- **Post-Processing:** Formats the final summary into bullet points and neatly wraps it into a paragraph.
|
| 27 |
+
- **Modular and Extensible:** Each step is modular, making it easy to adjust, extend, or integrate with other systems.
|
| 28 |
+
|
| 29 |
+
## Requirements
|
| 30 |
+
|
| 31 |
+
- Python 3.7+
|
| 32 |
+
- [spaCy](https://spacy.io/) with the `en_core_web_sm` model
|
| 33 |
+
- [NLTK](https://www.nltk.org/) (with the `punkt` tokenizer)
|
| 34 |
+
- [Transformers](https://huggingface.co/transformers/)
|
| 35 |
+
- [PyMuPDF](https://pymupdf.readthedocs.io/en/latest/)
|
| 36 |
+
- [BeautifulSoup4](https://www.crummy.com/software/BeautifulSoup/bs4/doc/)
|
| 37 |
+
|
| 38 |
+
## Installation
|
| 39 |
+
|
| 40 |
+
1. **Clone the repository:**
|
| 41 |
+
|
| 42 |
+
```bash
|
| 43 |
+
git clone https://github.com/yourusername/Medical_Doc_Summarization.git
|
| 44 |
+
cd Medical_Doc_Summarization
|
| 45 |
+
```
|