rigonsallauka
commited on
Commit
•
e9fe1e7
1
Parent(s):
7145780
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,66 @@
|
|
1 |
-
---
|
2 |
-
license: apache-2.0
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: apache-2.0
|
3 |
+
datasets:
|
4 |
+
- rigonsallauka/italian_ner_dataset
|
5 |
+
language:
|
6 |
+
- it
|
7 |
+
metrics:
|
8 |
+
- f1
|
9 |
+
- precision
|
10 |
+
- recall
|
11 |
+
- confusion_matrix
|
12 |
+
base_model:
|
13 |
+
- google-bert/bert-base-cased
|
14 |
+
pipeline_tag: token-classification
|
15 |
+
tags:
|
16 |
+
- NER
|
17 |
+
- medical
|
18 |
+
- symptom
|
19 |
+
- extraction
|
20 |
+
- italian
|
21 |
+
---
|
22 |
+
# Italian Medical NER
|
23 |
+
|
24 |
+
## Use
|
25 |
+
- **Primary Use Case**: This model is designed to extract medical entities such as symptoms, diagnostic tests, and treatments from clinical text in the Italian language.
|
26 |
+
- **Applications**: Suitable for healthcare professionals, clinical data analysis, and research into medical text processing.
|
27 |
+
- **Supported Entity Types**:
|
28 |
+
- `PROBLEM`: Diseases, symptoms, and medical conditions.
|
29 |
+
- `TEST`: Diagnostic procedures and laboratory tests.
|
30 |
+
- `TREATMENT`: Medications, therapies, and other medical interventions.
|
31 |
+
|
32 |
+
## Training Data
|
33 |
+
- **Data Sources**: Annotated datasets, including clinical data and translations of English medical text into Italian.
|
34 |
+
- **Data Augmentation**: The training dataset underwent data augmentation techniques to improve the model's ability to generalize to different text structures.
|
35 |
+
- **Dataset Split**:
|
36 |
+
- **Training Set**: 80%
|
37 |
+
- **Validation Set**: 10%
|
38 |
+
- **Test Set**: 10%
|
39 |
+
|
40 |
+
## Model Training
|
41 |
+
- **Training Configuration**:
|
42 |
+
- **Optimizer**: AdamW
|
43 |
+
- **Learning Rate**: 3e-5
|
44 |
+
- **Batch Size**: 64
|
45 |
+
- **Epochs**: 200
|
46 |
+
- **Loss Function**: Focal Loss to handle class imbalance
|
47 |
+
- **Frameworks**: PyTorch, Hugging Face Transformers, SimpleTransformers
|
48 |
+
|
49 |
+
## How to Use
|
50 |
+
You can easily use this model with the Hugging Face `transformers` library. Here's an example of how to load and use the model for inference:
|
51 |
+
|
52 |
+
```python
|
53 |
+
from transformers import AutoTokenizer, AutoModelForTokenClassification
|
54 |
+
import torch
|
55 |
+
|
56 |
+
model_name = "rigonsallauka/italian_medical_ner"
|
57 |
+
|
58 |
+
# Load the tokenizer and model
|
59 |
+
tokenizer = AutoTokenizer.from_pretrained(model_name)
|
60 |
+
model = AutoModelForTokenClassification.from_pretrained(model_name)
|
61 |
+
|
62 |
+
# Sample text for inference
|
63 |
+
text = "Il paziente ha lamentato forti mal di testa e nausea che persistevano da due giorni. Per alleviare i sintomi, gli è stato prescritto il paracetamolo e gli è stato consigliato di riposare e bere molti liquidi."
|
64 |
+
|
65 |
+
# Tokenize the input text
|
66 |
+
inputs = tokenizer(text, return_tensors="pt")
|