A Rust-based system that estimates neurobiological primitives (dopamine, serotonin, norepinephrine, adenosine, cortisol, glucose, circadian phase) from activity data using research-based computational models.

Takes activity data (sleep, exercise, meals, caffeine, stress, etc.) and computes biological state estimates based on neuroscience research. Instead of hardcoded impacts, it derives effects from event properties using peer-reviewed formulas.

# With your own data file
cargo run --bin cli -- /path/to/your_data.json

# With mock data
cargo run --bin cli -- mock_data.json

# Use a preset profile
cargo run --bin cli -- --profile chronically_sleep_deprived

cargo run --bin server

Then open http://localhost:8080 in your browser.

Pre-built profiles for testing:

• healthy - Well-rested, regular exercise, good nutrition
• sleep_deprived - Multiple nights of poor sleep
• shift_worker - Night shifts with circadian disruption
• athlete - Intense exercise and optimized schedule
• high_stress - Poor sleep, high stress, irregular meals

Input is JSON with events that have properties (not hardcoded impacts):

{
  "user_id": "user_123",
  "events": [
    {
      "event_id": "evt_1",
      "event_type": "sleep",
      "timestamp": "2025-01-15T23:00:00Z",
      "end_timestamp": "2025-01-16T07:00:00Z",
      "properties": {
        "duration_hours": 8.0,
        "quality": "good",
        "efficiency": 0.92
      }
    },
    {
      "event_id": "evt_2",
      "event_type": "caffeine",
      "timestamp": "2025-01-16T08:00:00Z",
      "properties": {
        "dose_mg": 100
      }
    },
    {
      "event_id": "evt_3",
      "event_type": "exercise",
      "timestamp": "2025-01-16T17:00:00Z",
      "properties": {
        "type": "cardio",
        "intensity": "moderate",
        "duration_minutes": 30
      }
    }
  ]
}

• sleep - Duration, quality, efficiency
• caffeine - Dose in mg
• exercise - Type, intensity, duration
• meal - Macros (protein/carbs/fats %), glycemic index
• light - Intensity in lux, wavelength
• stress - Severity, controllability, social evaluation
• social - Type (positive/negative), quality
• screen - Duration, content type, blue light
• nap - Duration
• interruption - Frequency

The system produces:

• Current state estimates for all 7 primitives
• Contributing events with computed impacts
• Detected patterns (sleep deprivation, withdrawal, synergies)
• Sleep drive analysis (homeostatic + circadian)
• Natural language interpretation

For natural language interpretation via OpenAI:

export OPENAI_API_KEY="your-key-here"

The system works without an API key; interpretation just won't be available.

src/
├── lib.rs         # Core estimation engine
├── chat.rs        # Natural language interpretation
├── profiles.rs    # Pre-built test profiles
├── bin/
│   ├── cli.rs     # Command-line interface
│   └── server.rs  # Web server demo
static/
├── index.html     # Web interface
└── chat.js        # Frontend JS

# Development
cargo build

# Optimized release
cargo build --release
./target/release/server

MIT