Skip to content

Latest commit

 

History

History
246 lines (193 loc) · 6.76 KB

README_new.md

File metadata and controls

246 lines (193 loc) · 6.76 KB

Bright Steps - Intelligent Child Development Assistant

Overview

Bright Steps is an innovative child development tracking system that combines modern AI techniques with expert knowledge to help parents and caregivers track, understand, and support their child's development journey. The system uses Retrieval-Augmented Generation (RAG) to provide personalized, evidence-based answers about child development.

Development Categories

Bright Steps tracks development across key areas defined by leading pediatric organizations:

🏃‍♂️ Motor Skills

  • Gross Motor: Large muscle movements like crawling, walking, running
  • Fine Motor: Small muscle control for tasks like grasping, drawing, using utensils

🗣️ Speech and Language

  • Receptive Language: Understanding words and directions
  • Expressive Language: Using words, gestures, and facial expressions
  • Speech Sounds: Pronunciation and articulation

👥 Social and Emotional

  • Social Interaction: Playing with others, sharing, turn-taking
  • Emotional Expression: Showing feelings, self-regulation
  • Attachment: Bonding with caregivers and forming relationships

🧠 Cognitive Development

  • Problem Solving: Finding solutions, understanding cause and effect
  • Memory: Recalling events, learning routines
  • Attention: Focusing on tasks, following directions
  • Early Learning: Colors, numbers, shapes, letters

🥄 Self-Help Skills

  • Feeding: Using utensils, drinking from cups
  • Dressing: Managing clothes, shoes, buttons
  • Toileting: Bathroom independence
  • Personal Care: Washing hands, brushing teeth

Key Features

Profile Management

  • 👶 Create and manage child profiles
  • 📊 Track age-appropriate milestones
  • 🏥 Record medical considerations
  • 🎯 Monitor focus areas

Development Tracking

  • ✅ Age-appropriate milestone checklists
  • 📈 Progress tracking over time
  • 🎨 Multiple development categories
  • 📝 Notes and observations

AI-Powered Assistance

  • 🧠 Intelligent question answering about child development
  • 📚 Evidence-based information retrieval
  • 🔍 Context-aware responses
  • 👤 Profile-aware suggestions

User Interface

  • 💻 Clean, intuitive Streamlit interface
  • 📱 Mobile-friendly design
  • 🔄 Real-time updates
  • 📊 Visual progress indicators

Getting Started

Prerequisites

  • Python 3.10 or higher
  • Virtual environment (recommended)
  • OpenAI API key
  • Pinecone API key (for vector storage)

Installation

  1. Clone the repository:
git clone [your-repo-url]
cd ragsteps
  1. Create and activate virtual environment:
# Windows
python -m venv rags
rags\Scripts\activate

# Linux/Mac
python -m venv rags
source rags/bin/activate
  1. Install dependencies:
pip install -r requirements.txt

API Setup

OpenAI Configuration

  1. Visit OpenAI API Keys
  2. Create a new API key
  3. Add to your .env file:
OPENAI_API_KEY=your_openai_api_key_here

Cost Considerations:

  • GPT-3.5-turbo: ~$0.002 per query
  • Embedding: ~$0.0001 per 1K tokens
  • Set monthly limits in OpenAI dashboard

Pinecone Setup

  1. Create account at Pinecone
  2. Create a new project
  3. Create an index:
    • Dimensions: 768 (DPR embedding size)
    • Metric: Cosine
    • Pod Type: p1.x1 (starter)
  4. Add to your .env:
PINECONE_API_KEY=your_pinecone_api_key_here
PINECONE_ENVIRONMENT=your_pinecone_environment
PINECONE_INDEX_NAME=your_index_name

Rate Limits:

  • Free tier: 1 request/second
  • Starter: 100 requests/second
  • Monitor usage in Pinecone dashboard

Running the Application

  1. Start the Streamlit interface:
streamlit run web/main_streamlit_with_profiles.py
  1. Open your browser:
  • The interface will automatically open at http://localhost:8501

Usage Guide

Creating a Profile

  1. Use the sidebar to create a new profile
  2. Enter child's name and date of birth
  3. Add any medical considerations
  4. Specify current focus areas

Tracking Development

  1. Select a profile to load
  2. View age-appropriate milestones
  3. Check off completed milestones
  4. Track progress over time

Getting Development Advice

  1. Ask questions about child development
  2. Receive personalized, evidence-based answers
  3. Get suggestions based on the child's profile

Technical Details

Components

  • Frontend: Streamlit
  • Embeddings: DPR (facebook/dpr-question_encoder-single-nq-base)
  • Vector Store: Pinecone
  • LLM: GPT-3.5-turbo
  • Profile Storage: Local JSON files

Project Structure

ragsteps/
├── web/                  # Web interface
│   └── main_streamlit_with_profiles.py
├── src/
│   ├── models/          # Data models
│   ├── rag/            # RAG implementation
│   └── utils/          # Utilities
├── data/               # Development data
└── user_data/         # User profiles

Data Sources

Bright Steps uses evidence-based developmental guidelines from leading organizations:

CDC Developmental Milestones

  • Age-specific milestones from 2 months to 5 years
  • Updated guidelines (2022) based on latest research
  • CDC Milestone Checklist

WHO Growth Standards

AAP Recommendations

Expert Resources

  • Zero to Three developmental guidelines
  • Child development research publications
  • Pediatric specialist recommendations

Security & Privacy

  • All profile data is stored locally
  • API keys are secured in .env
  • No personal data is sent to external services
  • Regular security updates

Deployment Options

Local Deployment

  1. Follow installation steps above
  2. Run with streamlit run
  3. Access via localhost

Docker Deployment

# Build image
docker build -t bright-steps .

# Run container
docker run -p 8501:8501 bright-steps

Troubleshooting

Common Issues

  1. Profile Not Saving

    • Check write permissions in user_data directory
    • Verify profile format

  2. RAG Not Responding

    • Verify API keys in .env
    • Check internet connection
    • Confirm Pinecone index is running

  3. Milestone Tracking

    • Ensure date of birth is correct
    • Check age calculations

Contributing

We welcome contributions! Please feel free to submit pull requests or open issues for discussion.

License

This project is licensed under the MIT License - see the LICENSE file for details.