Trend Visualization
7-day and 30-day trend charts for every tracked metric — built with fl_chart for smooth, responsive rendering on any screen size.
Movement Health
Your walking data, understood deeply — entirely on your device.
Movement Health is a mobile health intelligence application that turns the walking and activity data already being collected by your phone and wearables into meaningful, actionable insight. It connects to Apple HealthKit on iOS and Google Health Connect on Android, analyzes nine clinically relevant movement metrics, and surfaces patterns that would otherwise remain invisible in the noise of daily life.
What makes it unusual is where the analysis happens. All AI inference runs entirely on the device — using Apple Foundation Models on newer iPhones, Gemini Nano on capable Android flagships, and a user-selected GGUF model via llama.cpp as a universal fallback. Your health data never touches a remote server — not even ours.
The application is designed around clinical relevance. Metrics like walking asymmetry, double support time, and walking steadiness are established biomarkers for fall risk, neurodegenerative conditions, and cardiovascular health. Movement Health establishes your personal baseline over time, tracks trends across 7 and 30-day windows, and alerts you when patterns shift in ways that actually matter.
What it tracks
Nine movement metrics
Each pulled directly from HealthKit or Health Connect — no third-party data, no estimates.
Walking AsymmetryiOS only
Double Support Time
Step Length
Walking Speed
Step Count
VO₂ Max
Walking Heart Rate
Flights Climbed
Walking SteadinessiOS only
Architecture
On-device intelligence, no compromises
The core technical challenge was making meaningful AI analysis work on consumer hardware without network access. Movement Health solves this with a three-tier AI runtime: the app tries Apple Foundation Models first — zero download, GPU-accelerated, available on iOS 26+ — then falls back to Gemini Nano on supported Android devices, and as a universal fallback deploys llama.cpp with a user-provided GGUF model file.
The health context injected into the AI system prompt is carefully constrained to under 400 tokens — enough to be medically relevant, small enough to remain fast on edge hardware. A compact plain-text summary of your live metric snapshots and recent trends is built by a HealthContextBuilder and provided to the model at each session.
Environmental context from Open-Meteo — a no-API-key, privacy-respecting weather service — is also injected, making AI responses aware of conditions outside. A Pearson correlation engine runs on your historical data to surface relationships between metric pairs (for example, how your walking asymmetry varies with step count or sleep quality), cached locally in SQLite for seven days.
Technical Stack
- Flutter (Dart) iOS 15+ and Android 8.0+
- Apple HealthKit / Health Connect Native health data APIs
- Apple Foundation Models On-device AI, iOS 26+ priority
- Gemini Nano (ML Kit) On-device AI, Android flagships
- llama.cpp (GGUF) Universal on-device fallback
- SQLite Local analytics and correlation cache
- Riverpod State management
- CloudKit / Google Drive Optional encrypted backup
Capabilities
Built for the long view
Personalized Baselines
The app learns your normal ranges over time. Alerts fire when a metric moves meaningfully outside your personal baseline — not a generic population average.
Correlation Analysis
A Pearson correlation engine surfaces relationships between metric pairs — revealing patterns like how step count variation correlates with walking symmetry over time.
Smart Alerts
Configurable threshold alerts per metric, delivered via local notifications. Both deteriorations and improvements trigger — because improvement trends matter too.
Timed Walk Test
An in-app guided walk test with GPS distance tracking and live step counting — results saved to history for longitudinal comparison with healthcare providers.
Clinical PDF Reports
Generate a formatted PDF summary of your movement data — designed to be shared directly with a doctor, physical therapist, or specialist.
Coming Soon
Your health data, finally working for you.
Movement Health is in active development. The data your phone already collects every day is more valuable than most people realize — we're building the tool that makes it legible.