|
|
--- |
|
|
library_name: transformers |
|
|
tags: |
|
|
- medical |
|
|
license: mit |
|
|
language: |
|
|
- en |
|
|
base_model: |
|
|
- ShahRishi/OphthaBERT-v2 |
|
|
pipeline_tag: text-classification |
|
|
--- |
|
|
|
|
|
# OphtaBERT Glaucoma Classifier |
|
|
|
|
|
Binary classification for glaucoma diagnosis extraction from unstructured clinical notes. |
|
|
|
|
|
--- |
|
|
|
|
|
## Model Details |
|
|
|
|
|
### Model Description |
|
|
|
|
|
This model is a fine-tuned variant of OphthaBERT, which was pretrained on over 2 million clinical notes. It has been fine-tuned for binary classification on labeled clinical notes from Massachusetts Eye and Ear Infirmary. |
|
|
|
|
|
- **Finetuned from model:** [OphthaBERT-v2] |
|
|
|
|
|
--- |
|
|
|
|
|
## Uses |
|
|
|
|
|
We suggest utilizing this model in a zero-shot manner to generate binary glaucoma labels for each clinical note. For continued training on limited data, we recommend freezing the first 10 layers of the model. |
|
|
|
|
|
### Direct Use |
|
|
|
|
|
Use the code below to get started with the model: |
|
|
|
|
|
```python |
|
|
from transformers import AutoModelForSequenceClassification, AutoTokenizer |
|
|
|
|
|
# Load the fine-tuned model and tokenizer |
|
|
model = AutoModelForSequenceClassification.from_pretrained("ShahRishi/OphthaBERT-v2-glaucoma-binary") |
|
|
tokenizer = AutoTokenizer.from_pretrained("ShahRishi/OphthaBERT-v2-glaucoma-binary") |
|
|
|
|
|
# Example: Classify a clinical note |
|
|
clinical_note = "Example clinical note text..." |
|
|
inputs = tokenizer(clinical_note, return_tensors="pt", truncation=True, max_length=512) |
|
|
outputs = model(**inputs) |
|
|
|
