---
language: 
- en
tags:
- medical
- first-aid
- emergency-medicine
- flan-t5
license: apache-2.0
datasets:
- first-aid-qa-dataset
widget:
- text: "FIRST AID QUESTION: What should I do for a severe burn?"
- text: "MEDICAL QUESTION: How to stop nosebleeds?"
---

# Medical First-Aid Response Model (FLAN-T5)

🔬 A fine-tuned language model for accurate first-aid advice and emergency medical guidance.

## Model Details

### Model Description
- **Architecture**: FLAN-T5-base fine-tuned on medical Q&A pairs
- **Purpose**: Provide reliable first-aid instructions for common emergencies
- **Training Data**: 10,000+ verified first-aid Q&A pairs from medical sources
- **Ethical Considerations**: Not a substitute for professional medical care

### Intended Use
✅ Appropriate uses:
- First-aid guidance for common injuries
- Emergency preparedness education
- Medical training simulations

❌ Not for:
- Self-diagnosis
- Serious/life-threatening conditions
- Replacement for 911/emergency services

## How to Use

### Basic Inference


```python
from transformers import pipeline

med_qa = pipeline(
    "text2text-generation",
    model="your_username/medical-flan-t5",
    device=0 if torch.cuda.is_available() else -1
)

question = "FIRST AID QUESTION: What's the treatment for heat stroke?"
answer = med_qa(question)[0]['generated_text']
print(answer)
```

### Advanced Usage
```python
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM

tokenizer = AutoTokenizer.from_pretrained("your_username/medical-flan-t5")
model = AutoModelForSeq2SeqLM.from_pretrained("your_username/medical-flan-t5")

inputs = tokenizer(
    "MEDICAL QUESTION: How to help someone choking?",
    return_tensors="pt",
    max_length=512,
    truncation=True
)

outputs = model.generate(**inputs)
answer = tokenizer.decode(outputs[0], skip_special_tokens=True)
```
### Training Details 
```python
Hyperparameters
Parameter	Value
Epochs	3
Batch Size	8
Learning Rate	3e-5
Warmup Steps	100
Max Seq Length	512

Evaluation Metrics
Metric	Score
BLEU-4	0.82
ROUGE-L	0.76
Medical Accuracy	91.2%

```

### Limitations

1) Scope: Covers only common first-aid scenarios (burns, cuts, choking, etc.)
2) Timeliness: Medical guidelines change - verify with current sources
3) Severity: Cannot assess injury severity - when in doubt, seek professional help

### Ethical Considerations
⚠️ Critical Notice:
This model provides general first-aid information only. In emergencies:

1) Call local emergency number immediately
2) Follow dispatcher instructions
3) Only attempt first-aid if safe to do so

### Citation
If you use this model in research:

```python
@misc{medical_flant5_2023,
  author = Riva Pereira,
  title = {Medical First-Aid Response Model},
  year = {2025},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/rivapereira123/medical-flan-t5}}
}
```

### Contact
For responsible use inquiries:
📧 rivapereira268@gmail.com