JCholder/womens-health-chatbot

Women's Health Chatbot 🏥

This GPT-2 model has been fine-tuned to provide helpful, empathetic responses to questions about women's health and reproductive wellness.

Model Description

• Base Model: GPT-2 Medium
• Training Data: 8,278 conversation pairs from women's health Q&A
• Fine-tuning: Specialized for women's health topics
• Features: Context-aware responses, emergency detection, medical disclaimers

Topics Covered

• Menstrual health and cycle issues
• Sexual and reproductive health
• Pregnancy and fertility questions
• Gynecological conditions (PCOS, endometriosis, etc.)
• General women's wellness
• Emergency situation detection

Performance Metrics

• BLEU Score: 0.0057
• Perplexity: 135.97
• Domain Relevance: 0.71
• Training Loss: 0.7526

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

# Load the model
model = AutoModelForCausalLM.from_pretrained("JCholder/womens-health-chatbot")
tokenizer = AutoTokenizer.from_pretrained("JCholder/womens-health-chatbot")

def ask_health_question(question):
    prompt = f"USER: {question}\nDOCTOR:"
    inputs = tokenizer(prompt, return_tensors="pt", max_length=400, truncation=True)
    
    outputs = model.generate(
        **inputs,
        max_new_tokens=120,
        temperature=0.7,
        top_p=0.9,
        repetition_penalty=1.3,
        do_sample=True,
        pad_token_id=tokenizer.eos_token_id
    )
    
    full_response = tokenizer.decode(outputs[0], skip_special_tokens=True)
    response = full_response.split("DOCTOR:")[-1].strip()
    return response

# Example usage
question = "What causes irregular periods?"
response = ask_health_question(question)
print(response)

NextJS Integration

// NextJS API Route (pages/api/chat.js)
import { pipeline } from '@huggingface/transformers';

let chatbot;

async function initializeChatbot() {
  if (!chatbot) {
    chatbot = await pipeline('text-generation', 'JCholder/womens-health-chatbot');
  }
  return chatbot;
}

export default async function handler(req, res) {
  const { message } = req.body;
  const model = await initializeChatbot();
  
  const prompt = `USER: ${message}\nDOCTOR:`;
  const result = await model(prompt, {
    max_new_tokens: 100,
    temperature: 0.7,
    return_full_text: false
  });
  
  const response = result[0].generated_text.trim();
  res.json({ response });
}

Safety Features

• 🚨 Emergency keyword detection (suicide, severe bleeding, etc.)
• 🎯 Domain boundary enforcement (women's health only)
• 📋 Medical disclaimer integration
• 🛡️ Appropriate response filtering

Limitations

• This model is for educational and informational purposes only
• Not a substitute for professional medical advice
• Should not be used for emergency medical situations
• Always consult healthcare providers for serious concerns

Model Details

• Training Steps: 6207
• Parameters: ~354,823,168 parameters
• Framework: PyTorch, Transformers
• Fine-tuning Method: Causal Language Modeling

Ethical Considerations

This model has been designed with healthcare ethics in mind:

• Encourages professional medical consultation
• Avoids giving specific medical diagnoses
• Includes appropriate disclaimers
• Respects user privacy and autonomy

Citation

@misc{womens-health-chatbot,
  title={Women's Health Chatbot},
  author={Juliana Holder},
  year={2025},
  url={https://huggingface.co/JCholder/womens-health-chatbot}
}

Dataset

Trained on the altaidevorg/women-health-mini dataset containing professional women's health Q&A pairs.

Contact

For questions about this model, please refer to the project documentation or contact the model author.