Meal Log Insights - CGM Analysis PoC

A Streamlit-based proof of concept that analyzes meal images and their impact on glucose levels using Continuous Glucose Monitor (CGM) data.

🚀 Features

• Meal Image Analysis: Upload meal photos to automatically identify dishes and estimate nutritional content using Google's Gemini AI
• CGM Data Integration: Analyze glucose response patterns before and after meals
• Meal Impact Analysis: Calculate key metrics like glucose rise, time to peak, and return to baseline
• Reference Comparisons: Compare current meal impacts with similar past meals
• Editable Nutrition Data: Manually adjust identified dishes and nutritional estimates
• AI-Powered Insights: Get personalized recommendations based on your glucose response patterns

📋 Prerequisites

• Python 3.8 or higher
• Google Gemini API key (Get one here)
• CGM data in CSV format
• Food log data in CSV format

🛠️ Installation

1. Clone the repository:

git clone <repository-url>
cd Meal-Log-Insights

2. Install dependencies:

pip install -r requirements.txt

📊 Data Format

CGM Data CSV Format

user_id,date,time,value
100067072,"April 16, 2025",01:05,77.14

Food Logs CSV Format

user_id,date,time,slot,meal name,calories,protein,carbohydrates,fat,fibre
100067072,2025-04-17,08:57,BREAKFAST,"Masala Omelette, Wheat Roti",429.54,18.69,41.88,21.8,6.87

🚀 Running the Application

1. Start the Streamlit app:

streamlit run app.py

2. Open your browser to http://localhost:8501

📖 Usage

1. Configuration (Sidebar):

   • Enter your Gemini API key
   • Upload your CGM data CSV file
   • Upload your food logs CSV file
   • (Optional) Add a custom system prompt for AI insights
   • Enter your User ID

2. Meal Analysis:

   • Select the date and time of your meal
   • Choose the meal slot (Breakfast, Lunch, Dinner, Snack)
   • Upload a photo of your meal
   • Click "Analyze Meal"

3. Review Results:

   • View identified dishes and estimated nutrition
   • Check glucose metrics (pre-meal, peak, rise)
   • Read AI-generated insights and recommendations

4. Edit and Re-analyze (Optional):

   • Modify identified dishes or nutritional values
   • Click "Update & Re-analyze" for updated insights

🏗️ Project Structure

Meal-Log-Insights/
├── app.py                          # Main Streamlit application
├── requirements.txt                # Python dependencies
├── models.py                       # Pydantic models for data structures
├── data_processing/
│   ├── __init__.py
│   ├── cgm_data_processor.py      # CGM data loading and preprocessing
│   ├── metrics_processor.py        # Glucose metrics calculations
│   └── meal_impact_analysis.py    # Meal impact analysis
└── utils/
    └── llm_handler.py             # Gemini AI integration

🔧 Key Components

Metrics Calculated

• Pre-meal glucose: Average glucose 30 minutes before meal
• Post-meal peak: Maximum glucose within 3 hours after meal
• Time to peak: Minutes from meal to peak glucose
• Glucose rise: Difference between peak and pre-meal glucose
• Return to baseline: Whether glucose returned to pre-meal levels

Time Windows

• Pre-meal window: 30 minutes before meal
• Post-meal window: 3 hours after meal (2 hours for snacks)

⚠️ Important Notes

• This is a proof of concept and should not be used for medical decisions
• Always consult healthcare professionals for diabetes management
• Ensure your CSV files follow the exact format specified
• The system requires sufficient CGM data around meal times for accurate analysis

🤝 Contributing

Feel free to submit issues or pull requests to improve the functionality.

📄 License

This project is for demonstration purposes only.