metadata

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

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

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

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

Scope: Covers only common first-aid scenarios (burns, cuts, choking, etc.)
Timeliness: Medical guidelines change - verify with current sources
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:

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

Citation

If you use this model in research:

@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: 📧 [email protected]