Khyatimirani/pcos-clinical-supplements

---
license: apache-2.0
task_categories:
- question-answering
language:
- en
tags:
- medical
- diet
- supplements
- pcos
- women-health
size_categories:
- n<1K
---
### Dataset Overview
The dataset focuses on dietary supplements and lifestyle support for PCOS, based on evidence from clinical literature and structured knowledge extraction.

## It is intended for:

Fine-tuning conversational AI models
Building AI health assistants
Retrieval-augmented generation (RAG) systems
Research in digital health and women’s health AI
## 🎯 Objective

Women with PCOS often face confusion between doctor visits, relying on fragmented information from the internet or generic AI tools.

## This dataset aims to:

Bridge the “between consultation” care gap
Provide simple, clinically aligned guidance
Simulate empathetic, human-like conversations
Support continuous care through AI
📊 Dataset Structure

Each entry follows a chat template format:

`{
"messages": [
{"role": "user", "content": "..."},
{"role": "assistant", "content": "..."}
]
}`

## Key Characteristics:
Natural, non-medical language in user queries
Simple, clinically grounded responses
No unnecessary jargon
Focus on clarity and trust

## 🧩 Coverage

The dataset spans multiple aspects of PCOS:

Symptoms & Concerns
Irregular periods
Acne & hair issues
Weight gain
Fatigue
Mood swings
Fertility concerns
Core Mechanisms
Insulin resistance
Hormonal imbalance
Inflammation
Supplements Covered
Myo-inositol
Vitamin D
Vitamin E
Probiotics / Prebiotics
Omega-3 fatty acids
Zinc
Cinnamon
Selenium
CoQ10
Curcumin
Others (supportive evidence)

## User Intent Types
“Is this normal?”
“What should I take?”
“Can I avoid medicines?”
“What helps with fertility?”
“Why is this happening to me?”

## 🧠 Design Principles
1. Human-Centric Language

Questions reflect how real users speak — emotional, uncertain, and context-driven.

2. Clinically Aligned Responses

Answers are grounded in evidence, but simplified for accessibility.

3. No Overclaiming
Supplements are presented as supportive, not curative
Encourages balanced and informed decision-making
4. Personalization-Oriented

## Content is structured to enable:

Symptom → recommendation mapping
Context-aware responses
⚕️ Safety & Limitations
This dataset is for informational purposes only
Not a substitute for medical advice
Does not include dosage recommendations
Does not handle emergency or critical cases

Models trained on this dataset should:

Avoid definitive diagnoses
Encourage professional consultation when needed

## 📚 Data Source

The dataset is derived from:

Clinical review literature on PCOS and dietary supplements
Evidence-based insights on metabolism, hormones, and fertility

# Primary reference used:

Link to paper: https://pmc.ncbi.nlm.nih.gov/articles/PMC11466749/

Dietary supplements in polycystic ovary syndrome – current evidence

Additional structuring and transformation performed to:

Convert clinical knowledge into conversational format
Ensure accessibility and clarity
🛠️ Use Cases
AI health companion for women’s health
PCOS education tools
Conversational AI research
Preventive healthcare applications
Symptom-aware recommendation systems

## 🚀 Future Improvements
Multi-turn conversations (context retention)
Personalization based on age, goals, and biomarkers
Integration with doctor-approved clinical pathways
Multilingual support (e.g., Hindi, regional languages)
Voice-based interaction datasets

## 🤝 Contribution

If you are:

A clinician
A researcher
A builder in women’s health

## We welcome collaboration to:

Improve clinical depth
Expand dataset coverage
Build safer and more effective AI systems

## 📬 Contact

For collaboration or queries, feel free to reach out.
[e-mail]

## ❤️ Motivation

Healthcare often begins and ends at consultations.
But for many women, especially in PCOS,
most of the journey happens in between.

This dataset is a step toward supporting that journey.