SweetiePy - Type 1 Diabetes Data Analysis

A Python package for accessing and analyzing Type 1 diabetes data from a DIY Loop system stored in MongoDB Atlas. Focuses on CGM (Continuous Glucose Monitor) pattern analysis for diabetes management optimization.

Features

• 📊 CGM Data Analysis - Access and analyze continuous glucose monitor data
• 💉 Pump Data Integration - Query insulin doses, basal rates, and treatment data
• 🔗 Merged Data Analysis - Synchronize CGM readings with active pump settings at each timestamp
• 📈 Time-Series Analysis - Built-in time-in-range calculations and statistics
• 🎯 Settings Correlation - Analyze how basal rates, carb ratios, and ISF affect glucose outcomes
• 🔍 Flexible Queries - Predefined periods or custom date ranges
• 🚀 High Performance - PyArrow-backed DataFrames for efficient processing
• 🔐 Secure - Environment-based configuration for credentials

Project Structure

src/sweetiepy/
├── connection/    # Database connectivity
│   └── mongodb.py
├── data/          # Data access modules
│   ├── cgm.py     # CGM data queries
│   ├── pump.py    # Pump/treatment data queries
│   └── merged.py  # CGM + pump settings synchronization
└── utils/         # Utilities and debugging
    └── debug.py
docs/              # Analysis documentation
dev/               # Development and analysis scripts
├── exploratory/   # Exploratory analysis notebooks
├── reports/       # Analysis reports  
├── usage_example.py # Complete usage demonstration
└── merged_data_example.py # CGM + pump settings analysis example
tests/             # Test modules

Installation

Recommended: Install with uv (fast, modern Python package manager)

# Install uv first if you don't have it
curl -LsSf https://astral.sh/uv/install.sh | sh

# Install sweetiepy
uv add sweetiepy

# Or install directly without adding to project
uv tool install sweetiepy

Alternative: Install with pip

# Install the latest version from PyPI
pip install sweetiepy

Install from Source (for development)

# Clone the repository
git clone <repository-url>
cd sweetiepy

# Install dependencies using uv (recommended)
uv sync

# Install package in editable mode for development
uv pip install -e .

Quick Start

1. Set Up MongoDB Connection

Create a .env file in your project directory with your MongoDB Atlas credentials:

MONGODB_USERNAME=your_actual_username
MONGODB_PW=your_actual_password
MONGODB_URI=mongodb+srv://<username>:<password>@cluster0.yourcluster.mongodb.net/?retryWrites=true&w=majority&appName=Cluster0
MONGODB_DATABASE=myCGMitc

Important Notes:

• Keep the <username> and <password> placeholders in the URI exactly as shown - the code automatically replaces them
• Only change the cluster URL part (after the @ symbol) to match your MongoDB Atlas cluster
• Provide your actual username and password in the separate MONGODB_USERNAME and MONGODB_PW variables
• Ensure the database user has read access to the myCGMitc database

2. Basic Usage

from sweetiepy.data.cgm import CGMDataAccess
from sweetiepy.data.pump import PumpDataAccess
from sweetiepy.data.merged import MergedDataAccess
from datetime import datetime, timedelta

# Initialize CGM data access
cgm = CGMDataAccess()
cgm.connect()

# Get last week's data as a cleaned DataFrame
df = cgm.get_dataframe_for_period('last_week')

# Basic analysis
analysis = cgm.analyze_dataframe(df)
print(f"Average glucose: {analysis['basic_stats']['avg_glucose']:.1f} mg/dL")
print(f"Time in range: {analysis['time_in_range']['normal_percent']:.1f}%")

# Get pump data (insulin treatments)
pump = PumpDataAccess()
pump.connect()
treatments = pump.get_dataframe_for_period('last_24h')

cgm.disconnect()
pump.disconnect()

Advanced: Merged CGM and Pump Settings Analysis

For advanced analysis that correlates CGM readings with active pump settings:

from sweetiepy.data.merged import MergedDataAccess

# Get CGM data synchronized with active pump settings
with MergedDataAccess() as merged:
    # Each CGM reading includes the pump settings active at that time
    df = merged.get_merged_cgm_and_settings(days=7)
    
    # Now you can analyze how settings affect glucose outcomes
    print(f"Data shape: {df.shape}")
    print(f"Columns: {list(df.columns)}")
    
    # Example: Average glucose by basal rate
    glucose_by_basal = df.groupby('active_basal')['sgv'].agg(['mean', 'count'])
    print("\nAverage Glucose by Basal Rate:")
    print(glucose_by_basal)
    
    # Correlation analysis
    analysis = merged.analyze_settings_correlation(df)
    if 'correlations' in analysis:
        print("\nCorrelations with glucose:")
        for setting, corr in analysis['correlations'].items():
            print(f"  {setting}: {corr:+.3f}")

3. Test Your Connection

For development installations, you can test the connection:

# If installed from source with uv
uv run python -m sweetiepy.connection.mongodb

# If installed via pip
python -m sweetiepy.connection.mongodb

You should see output like:

✓ Connected to MongoDB database: myCGMitc
Available databases: ['myCGMitc', 'test', 'admin', 'local']
Collections in myCGMitc: ['entries', 'treatments', 'food', 'settings', 'devicestatus', 'auth_roles', 'auth_subjects', 'activity', 'profile']
✓ Disconnected from MongoDB

Database Schema

The myCGMitc database contains the following collections:

• entries - CGM/blood glucose readings (primary data for analysis)
• treatments - Insulin doses and medical treatments
• food - Food intake and carbohydrate data
• settings - Loop system configuration
• devicestatus - Device status and connectivity info
• profile - User profile and basal rate settings
• activity - Activity and exercise logs
• auth_roles, auth_subjects - Authentication data

entries Collection Schema (CGM Data)

Collection Stats:

• Total documents: 243,047 CGM readings
• Date range: March 2023 to July 2025 (~2 years of data)
• Device: Dexcom CGM ("share2")
• Data actively updated (real-time)

Document Structure:

{
  "_id": "ObjectId",
  "sgv": 163,                           // Blood glucose value (mg/dL)
  "date": 1678724324000.0,             // Unix timestamp (milliseconds)
  "dateString": "2023-03-13T16:18:44.000Z",  // ISO formatted date
  "trend": 4,                          // Glucose trend indicator (1-7)
  "direction": "Flat",                 // Trend direction text
  "device": "share2",                  // CGM device identifier
  "type": "sgv",                       // Sensor glucose value type
  "utcOffset": 0,                      // UTC offset
  "sysTime": "2023-03-13T16:18:44.000Z"  // System timestamp
}

Key Fields:

• sgv - Primary glucose reading in mg/dL
• date - Unix timestamp for sorting and time-based queries
• direction - Trend indicators: "Flat", "FortyFiveUp", "FortyFiveDown", "SingleUp", "SingleDown", "DoubleUp", "DoubleDown"
• trend - Numeric trend value (1-7 scale)

Database Indexes:

• Optimized indexes on date, sgv, dateString, type for efficient queries

Development Stages

✅ Stage 1: Database Connection

• MongoDB Atlas connection with environment variables
• Basic authentication and connection testing
• Database and collection discovery

✅ Stage 2: CGM Data Access

• Connect to entries collection for glucose readings
• Explore document structure and schema
• Implement basic data retrieval queries
• Verify data format and field analysis

✅ Stage 3: Time-Range Queries

• Implement date/time filtering for CGM data
• Add functions to query specific time periods (24h, week, month, custom ranges)
• Test with various time ranges and validate results
• Summary statistics and data validation

✅ Stage 4: Data Processing & DataFrame Integration - COMPLETE

• MongoDB to pandas DataFrames with PyArrow backend
• Efficient data cleaning and validation
• Timestamp conversions and timezone management
• Time-series analysis preparation

✅ Stage 5: Pump Data Access - COMPLETE

• Access to insulin treatment data
• Bolus and basal rate queries
• Treatment type filtering
• Insulin and carbohydrate calculations

🔄 Current Focus: Pattern Analysis & Insights

• Time-of-day glucose patterns
• Treatment correlation analysis
• Weekly and monthly trends
• Statistical summaries by time period

Key Files & Commands

Core Modules:

• src/loopy/connection/mongodb.py - Main MongoDB connection module
• src/loopy/data/cgm.py - CGM data access and time-range queries
• src/loopy/utils/debug.py - Connection debugging utilities
• docs/analysis_patterns.md - Analysis methodology documentation
• .env.example - Environment variable template
• CLAUDE.md - Development guidance for AI assistants

Module Testing Commands:

Recommended with uv:

# Test database connection
uv run python -m sweetiepy.connection.mongodb

# Test CGM data access and time-range queries  
uv run python -m sweetiepy.data.cgm

# Test pump data access
uv run python -m sweetiepy.data.pump

# Debug connection issues
uv run python -m sweetiepy.utils.debug

With pip installations:

# Test database connection
python -m sweetiepy.connection.mongodb

# Test CGM data access
python -m sweetiepy.data.cgm

# Test pump data access
python -m sweetiepy.data.pump

# Debug connection issues
python -m sweetiepy.utils.debug

Development commands (for source installations):

# Start marimo notebook for exploration
uv run marimo edit dev/exploratory/analysis.py

# Run usage example (3 months of CGM data)
uv run python dev/usage_example.py

Troubleshooting

If you encounter authentication errors:

1. Verify credentials in MongoDB Atlas dashboard
2. Ensure database user has appropriate permissions
3. Check for extra spaces in .env file
4. Test connection with MongoDB Compass first
5. Run uv run python -m sweetiepy.utils.debug for detailed diagnostics

Development Standards

This project follows Python best practices for professional coding:

• Code Quality: Type hints, comprehensive docstrings, error handling
• Reproducibility: Pinned dependencies, environment configuration, deterministic workflows
• Documentation: Detailed docstrings, inline comments, complete setup instructions
• Testing: Input validation, data quality checks, comprehensive testing

Available Data Access Modules

CGM Data (loopy.data.cgm.CGMDataAccess)

• Access continuous glucose monitor readings
• Query by time periods: 'last_24h', 'last_week', 'last_month', 'last_3_months'
• Get pandas DataFrames with PyArrow backend for efficient analysis
• Built-in statistical analysis and time-in-range calculations

Pump Data (loopy.data.pump.PumpDataAccess)

• Access insulin pump treatment data
• Query bolus doses, basal rates, and temporary basals
• Filter by treatment types and time periods
• Calculate insulin on board (IOB) and carbohydrate data

Available Event Types

The pump data includes these treatment event types:

• Correction Bolus - Insulin bolus doses
• Carb Correction - Carbohydrate entries
• Temp Basal - Temporary basal rate adjustments
• Site Change - Pump site/pod changes
• Suspend Pump - Pump suspension events
• Temporary Override - Temporary setting overrides

Key Pump Methods

from sweetiepy.data.pump import PumpDataAccess

with PumpDataAccess() as pump:
    # Get specific treatment types
    boluses = pump.get_bolus_data(days=7)           # Bolus doses
    basals = pump.get_basal_data(days=7)            # Temp basals
    carbs = pump.get_carb_data(days=7)              # Carb entries
    site_changes = pump.get_site_change_data(days=30)  # Site changes
    
    # Get current pump status
    iob = pump.get_insulin_on_board()               # Current IOB
    cob = pump.get_carbs_on_board()                 # Current COB
    current_basal = pump.get_current_basal_rate()   # Current basal rate
    
    # Get pump profiles/settings
    basal_profile = pump.get_basal_profile()        # Basal rate schedule
    carb_ratios = pump.get_carb_ratio_profile()     # I:C ratios
    sensitivity = pump.get_insulin_sensitivity_profile()  # ISF values
    
    # Get DataFrame with all treatments
    df = pump.get_dataframe_for_period('last_week')
    analysis = pump.analyze_treatments(df)          # Summary statistics

Advanced Usage Examples

Custom Time Range Queries

from sweetiepy.data.cgm import CGMDataAccess
from datetime import datetime, timedelta

# Initialize and connect
cgm = CGMDataAccess()
cgm.connect()

# Query specific date range
start_date = datetime(2024, 1, 1)
end_date = datetime(2024, 1, 31)
january_data = cgm.get_readings_in_range(start_date, end_date)

# Get DataFrame with analysis
df = cgm.to_dataframe(january_data)
analysis = cgm.analyze_dataframe(df)

print(f"January 2024 Statistics:")
print(f"  Average: {analysis['basic_stats']['avg_glucose']:.1f} mg/dL")
print(f"  Std Dev: {analysis['basic_stats']['std_glucose']:.1f}")
print(f"  Time in Range (70-180): {analysis['time_in_range']['normal_percent']:.1f}%")
print(f"  Time High (>180): {analysis['time_in_range']['high_percent']:.1f}%")
print(f"  Time Low (<70): {analysis['time_in_range']['low_percent']:.1f}%")

cgm.disconnect()

Correlating CGM and Pump Data

from sweetiepy.data.cgm import CGMDataAccess
from sweetiepy.data.pump import PumpDataAccess
import pandas as pd

# Initialize both data access objects
cgm = CGMDataAccess()
pump = PumpDataAccess()

cgm.connect()
pump.connect()

# Get data for the same period
period = 'last_week'
cgm_df = cgm.get_dataframe_for_period(period)
pump_df = pump.get_dataframe_for_period(period)

# Filter for bolus doses only
bolus_df = pump_df[pump_df['eventType'] == 'Bolus']

print(f"Week Summary:")
print(f"  CGM Readings: {len(cgm_df)}")
print(f"  Total Boluses: {len(bolus_df)}")
print(f"  Total Insulin: {bolus_df['insulin'].sum():.1f} units")
print(f"  Average Bolus: {bolus_df['insulin'].mean():.2f} units")

cgm.disconnect()
pump.disconnect()

Pattern Analysis by Time of Day

from sweetiepy.data.cgm import CGMDataAccess
import pandas as pd

cgm = CGMDataAccess()
cgm.connect()

# Get a month of data
df = cgm.get_dataframe_for_period('last_month')

# Add hour of day
df['hour'] = df['dateTime'].dt.hour

# Calculate hourly statistics
hourly_stats = df.groupby('hour')['sgv'].agg(['mean', 'std', 'count'])
hourly_stats.columns = ['avg_glucose', 'std_dev', 'num_readings']

print("Hourly Glucose Patterns:")
print(hourly_stats.round(1))

# Find problematic times
high_times = hourly_stats[hourly_stats['avg_glucose'] > 180]
if not high_times.empty:
    print(f"\nHours with average glucose > 180 mg/dL:")
    for hour in high_times.index:
        print(f"  {hour:02d}:00 - Avg: {high_times.loc[hour, 'avg_glucose']:.1f}")

cgm.disconnect()

Pump Data Analysis Examples

Daily Insulin and Carb Summary

from sweetiepy.data.pump import PumpDataAccess
import pandas as pd
from datetime import datetime, timedelta

with PumpDataAccess() as pump:
    # Get a week of pump data
    df = pump.get_dataframe_for_period('last_week')
    
    # Analyze treatments
    analysis = pump.analyze_treatments(df)
    
    print("Week Summary:")
    print(f"  Total treatments: {analysis['total_treatments']}")
    print(f"  Date range: {analysis['date_range']['days_span']} days")
    
    if 'insulin_summary' in analysis:
        print(f"\nInsulin Summary:")
        print(f"  Total: {analysis['insulin_summary']['total_insulin']:.1f} units")
        print(f"  Daily average: {analysis['insulin_summary']['total_insulin']/7:.1f} units/day")
        print(f"  Average dose: {analysis['insulin_summary']['avg_dose']:.2f} units")
        print(f"  Range: {analysis['insulin_summary']['min_dose']:.2f} - {analysis['insulin_summary']['max_dose']:.2f} units")
    
    if 'carb_summary' in analysis:
        print(f"\nCarb Summary:")
        print(f"  Total: {analysis['carb_summary']['total_carbs']:.0f}g")
        print(f"  Daily average: {analysis['carb_summary']['total_carbs']/7:.0f}g/day")
        print(f"  Average per entry: {analysis['carb_summary']['avg_carbs']:.1f}g")

Insulin Patterns by Time of Day

from sweetiepy.data.pump import PumpDataAccess
import pandas as pd

with PumpDataAccess() as pump:
    # Get bolus data for the last month
    boluses = pump.get_bolus_data(days=30)
    
    if boluses:
        # Convert to DataFrame
        bolus_df = pd.DataFrame(boluses)
        bolus_df['dateTime'] = pd.to_datetime(bolus_df['timestamp'])
        bolus_df['hour'] = bolus_df['dateTime'].dt.hour
        
        # Calculate hourly insulin patterns
        hourly_insulin = bolus_df.groupby('hour')['insulin'].agg(['sum', 'mean', 'count'])
        hourly_insulin.columns = ['total_insulin', 'avg_bolus', 'num_boluses']
        
        print("Hourly Insulin Patterns (last 30 days):")
        print(hourly_insulin.round(2))
        
        # Find peak insulin times
        peak_hours = hourly_insulin.nlargest(3, 'total_insulin')
        print(f"\nPeak insulin hours:")
        for hour in peak_hours.index:
            print(f"  {hour:02d}:00 - Total: {peak_hours.loc[hour, 'total_insulin']:.1f}u, "
                  f"Avg: {peak_hours.loc[hour, 'avg_bolus']:.2f}u")

Pump Settings Review

from sweetiepy.data.pump import PumpDataAccess

with PumpDataAccess() as pump:
    # Get current pump settings
    basal_profile = pump.get_basal_profile()
    carb_ratios = pump.get_carb_ratio_profile()
    isf = pump.get_insulin_sensitivity_profile()
    
    print("Current Pump Settings:")
    
    print("\nBasal Profile:")
    total_basal = 0
    for i, entry in enumerate(basal_profile):
        print(f"  {entry['time']}: {entry['value']} U/hr")
        # Calculate duration until next entry
        if i < len(basal_profile) - 1:
            current_time = pd.to_datetime(f"2024-01-01 {entry['time']}")
            next_time = pd.to_datetime(f"2024-01-01 {basal_profile[i+1]['time']}")
            hours = (next_time - current_time).seconds / 3600
        else:
            # Last entry goes until midnight
            current_time = pd.to_datetime(f"2024-01-01 {entry['time']}")
            hours = (pd.to_datetime("2024-01-02 00:00") - current_time).seconds / 3600
        total_basal += entry['value'] * hours
    
    print(f"  Total daily basal: {total_basal:.1f} units")
    
    print("\nCarb Ratios (I:C):")
    for entry in carb_ratios:
        print(f"  {entry['time']}: 1:{entry['value']}g")
    
    print("\nInsulin Sensitivity Factors (ISF):")
    for entry in isf:
        print(f"  {entry['time']}: {entry['value']} mg/dL per unit")
    
    # Get current status
    iob = pump.get_insulin_on_board()
    cob = pump.get_carbs_on_board()
    print(f"\nCurrent Status:")
    print(f"  IOB: {iob:.2f} units")
    print(f"  COB: {cob:.1f}g")

Package Information

• PyPI: sweetiepy
• License: MIT
• Python: 3.12+
• Dependencies: pymongo, pandas, pyarrow, python-dotenv, python-dateutil

Security Notes

• Never commit the .env file to version control
• Use read-only database connections when possible
• Store connection credentials securely
• The package does not store or transmit credentials

Contributing

Contributions are welcome! This is an open-source project aimed at helping the Type 1 diabetes community better understand and manage their condition through data analysis.

Support

For issues, questions, or contributions, please visit the GitHub repository.