Miruns Connect

Warning

Early-stage open-source project — not for production use. This software is under active development and is provided as-is, without warranty of any kind. Expect breaking changes, incomplete features, and potential instability. Do not rely on it in any critical or production environment.

Not medical advice. Miruns Connect is a wellness and sport-science tool intended for informational and research purposes only. It is not a medical device, and nothing in this application should be interpreted as medical advice, diagnosis, or treatment. Always consult a qualified healthcare professional for any health-related decisions.

AI-assisted development. Portions of this codebase, including data-processing logic and application code were developed with the assistance of AI tools. Outputs have been reviewed by human developers, but no guarantee of correctness or completeness is implied at this stage.

Neuroscience meets sport. Miruns listens to your body so you can push further, safer.

Miruns is a cutting-edge platform at the intersection of neuroscience, sport science, and active living. Designed to integrate discreetly into everyday headphones, Miruns captures a comprehensive array of physiological and environmental biodata — without interrupting your training or your day.

The companion Flutter app processes signals from the embedded hardware and device sensors with AI to deliver four core pillars: real-time fatigue prediction, recuperation tracking, personalised cognitive health insights, and a first-person daily narrative — actionable awareness written by your body, for you.

Screenshots

More screenshots

Why Miruns?

Elite sport has always demanded an edge. Miruns brings neuroscience-grade biosignal monitoring to every athlete and active person — discreetly embedded in the headphones you already wear.

EEG-Powered Fatigue Intelligence

EEG signals provide a direct, objective window into brain activity — the most accurate method available for detecting mental fatigue. Miruns reads that signal continuously and turns it into four practical benefits:

Benefit	What it means for you
Accurate fatigue assessment	Objective, brain-level measurement of cognitive state — not a proxy like typing speed or eye tracking
Real-time prediction & alerts	Timely prompts to take a break, adjust intensity, or switch tasks before performance degrades
Personalised management	Individual fatigue patterns analysed over time → tailored recommendations for breaks, recovery, and work rhythm
Long-term cognitive health	Longitudinal tracking surfaces trends that may signal burnout risk or cognitive decline — before they become a problem

Most health apps show dashboards of numbers. Miruns goes further: it presents your biometrics as a narrative. Story-framing surfaces correlations you'd otherwise miss — poor sleep preceding an elevated resting heart rate, high AQI days correlating with fewer steps, or cognitive fatigue building through an intense interval session.

Who Benefits?

Sweat. Gain. Do.

Audience	Use case
Athletes & fitness enthusiasts	Optimal workout planning, overtraining prevention, post-exercise mental recovery tracking
Professionals & executives	Monitor fatigue across long working hours; optimise focus intervals and break timing for sustained productivity
Students	Identify when cognitive capacity is falling during study sessions; adjust pacing for better retention
Gamers & esports competitors	Gauge mental fatigue during marathon sessions; protect reaction time and decision-making under pressure
Drivers & operators	Long-haul truck drivers, pilots, and shift workers get continuous fatigue monitoring to reduce accident risk
Medical & clinical	Healthcare providers and researchers gain an objective tool for patient fatigue monitoring and fatigue-management studies

---

Features

Feature	Description
AI Journal	Daily narrative generated from real sensor data — headline, mood, summary, full body text. Written in first-person ("your body speaking to you").
Smart Refresh	Persisted entries return instantly. AI runs only when new unprocessed captures exist. No redundant sensor reads or API calls.
Background Captures	WorkManager-based periodic data collection with quiet-hour and battery-awareness support.
Manual Capture	On-demand snapshot with toggleable data sources (health, environment, location, calendar, BLE HR device).
BLE Heart Rate	Real-time streaming from any Bluetooth Low Energy Heart Rate Profile (0x180D) device — Polar H10, Wahoo TICKR, Garmin straps, etc. Live ECG-style waveform in the Capture screen. Continuous session recording accumulates timestamped BPM samples and raw RR intervals for the full connection duration; RMSSD, SDNN, and mean-RR are computed at capture time and stored alongside the session.
HRV & Autonomic Tone	RMSSD / SDNN computed from BLE RR intervals at every capture. A plain-language stress hint and a BPM arc narrative are generated and passed to the AI so it can interpret cardiac state as a story, not just a number.
Cardiovascular Depth	Resting heart rate and HRV (SDNN) also read from HealthKit / Health Connect alongside average HR. All three are included in daily snapshots and AI prompts.
Patterns & Trends	AI-derived insights aggregated from capture history — energy distribution, recurring themes, notable signals, recent moments timeline.
Nutrition Scanner	Scan any food barcode from the Capture screen. Open Food Facts returns Nutri-Score, NOVA group, and full macros. AI prompt includes the food data and generates a nutrition Context analysis correlating nutrition to HRV/energy patterns.
User Annotations	Free-text notes and mood emojis per day, persisted in SQLite alongside the AI-generated content.
Onboarding	Step-by-step permission flow with per-permission explanations and privacy notes. Every step is skippable.
BCI Signal Analysis	Four visualisation modes on the Live Signal screen: time-domain waveforms, real-time FFT spectral analysis (spectrum / waterfall / EEG bands), neural-state decoding demo (5-state classifier with confidence ring), and signal quality monitoring demo (SNR, impedance, artifact detection). Switchable via a popup menu — works with real BLE hardware and demo mode.
Dark & Light Themes	Material 3 theming with system-mode detection.

---

Data Sources

Domain	Sensor / API	What's read
Movement	Health Connect / HealthKit	Steps, distance, calories, workouts
Cardiovascular	Optical HR sensor (OS health store)	Average HR, resting HR, HRV (SDNN)
Cardiovascular	BLE HR strap (direct, 0x180D)	Continuous BPM session + RR intervals → RMSSD / SDNN / mean-RR; full BleHrSession stored per capture
Signal sources	BLE plugin system (ADS1299, …)	Multi-channel streaming from community hardware — 4 view modes (waveform, spectral, decoding, monitoring). See CONTRIBUTING_SOURCES.md
Sleep	Device sleep tracking	Duration, phases
Environment	GPS → ambient-scan API	Temperature, humidity, AQI, UV, pressure, wind, conditions
Schedule	Device calendar (CalDAV)	Today's events
Location	GPS (Geolocator)	Coordinates — ephemeral, never stored
Nutrition	Barcode → Open Food Facts API v2	Product name, Nutri-Score, NOVA group, macros per 100 g / per serving

Privacy: GPS is read once to fetch environmental data, then discarded. No location history is recorded or transmitted. Health data is read-only — the app never writes to HealthKit or Health Connect.

---

Tech Stack

Concern	Library	Role
State management	flutter_riverpod ^2.6.1	DI + reactive state
Routing	go_router ^14.6.2	Declarative navigation (3-tab shell + standalone routes)
Health	health ^11.1.0	Cross-platform HealthKit / Health Connect (HR, resting HR, HRV, steps…)
BLE	flutter_blue_plus ^1.35.3	BLE scan / connect / notify — Heart Rate Profile 0x180D
Location	geolocator ^13.0.2	GPS coordinates
Calendar	device_calendar ^4.3.2	CalDAV read
HTTP	http ^1.2.2	Ambient-scan & AI API calls
Barcode scanner	mobile_scanner ^7.0.1	Camera-based barcode reading (EAN-13, UPC-A, etc.)
Nutrition API	Open Food Facts API v2	Product lookup by barcode — no API key needed
Persistence	sqflite ^2.3.3 + path	Local SQLite (entries, captures, settings) — schema v11
Background	workmanager ^0.9.0	Periodic background captures
Notifications	flutter_local_notifications ^17.0.0	Daily reminders
Sharing	share_plus ^10.1.0	Share journal entries
Typography	google_fonts ^6.2.1	Inter (body) / Playfair Display (headlines)
Formatting	intl ^0.19.0	Date/number formatting
Config	flutter_dotenv ^6.0.0	.env file support

---

Getting Started

Prerequisites

• Flutter SDK ≥ 3.9.2
• iOS: Xcode 14+, deployment target iOS 14+
• Android: SDK 21+ (Android 5.0), Health Connect app for health data on Android 13+

Install & run

flutter pub get
flutter run

# Or use the helper scripts
.\dev.ps1           # Windows (PowerShell)
./dev.sh            # macOS / Linux

Environment

Copy .env.example to .env and fill in your API key:

AI_API_KEY=your_key_here

Platform configuration

iOS

Info.plist keys are pre-configured:

• NSLocationWhenInUseUsageDescription
• NSLocationAlwaysAndWhenInUseUsageDescription
• NSHealthShareUsageDescription / NSHealthUpdateUsageDescription
• NSCalendarsUsageDescription
• NSMotionUsageDescription
• NSBluetoothAlwaysUsageDescription ← BLE heart rate devices
• NSBluetoothPeripheralUsageDescription

Enable HealthKit capability in Xcode: Runner target → Signing & Capabilities → + HealthKit.

Android

AndroidManifest.xml includes permissions for location, Health Connect (including READ_HEART_RATE_VARIABILITY), activity recognition, calendar, and BLE:

• BLUETOOTH / BLUETOOTH_ADMIN (API ≤ 30 legacy)
• BLUETOOTH_SCAN (neverForLocation) + BLUETOOTH_CONNECT (API 31+)
• android.hardware.bluetooth_le feature declaration (required=false — app installs on non-BLE devices)
• CAMERA permission + android.hardware.camera / android.hardware.camera.autofocus features (required=false) — barcode scanner

The Health Connect app is required on Android 13+ for HR, resting HR, and HRV data.

---

Architecture

lib/
├── core/
│   ├── background/           # WorkManager callback (capture_executor.dart)
│   ├── models/               # Immutable domain objects
│   │   ├── body_blog_entry   #   BodyBlogEntry + BodySnapshot
│   │   ├── capture_entry     #   CaptureEntry + health/env/location sub-models
│   │   ├── capture_ai_metadata  # AI-derived tags, themes, energy level per capture
│   │   ├── nutrition_log     #   NutritionLog + NutritionFacts (barcode scanner)
│   │   ├── ai_models         #   ChatMessage, request/response DTOs
│   │   ├── ai_provider_config  # Provider presets & config model
│   │   └── background_capture_config
│   ├── router/               # GoRouter (3-tab shell + standalone routes)
│   ├── services/             # All business logic
│   │   ├── service_providers #   ★ Central Riverpod provider registry
│   │   ├── body_blog_service #   Smart-refresh orchestrator
│   │   ├── journal_ai_service  # Prompt → AI → JournalAiResult
│   │   ├── capture_metadata_service  # Per-capture background AI metadata
│   │   ├── ai_service        #   HTTP client for any OpenAI-compatible endpoint
│   │   ├── ai_config_service #   Persist & manage active AI provider
│   │   ├── local_db_service  #   SQLite CRUD (schema v11)
│   │   ├── capture_service   #   Multi-source data → CaptureEntry
│   │   ├── nutrition_service #   Open Food Facts API v2 (barcode lookup)
│   │   ├── background_capture_service  # WorkManager scheduler
│   │   ├── context_window_service  # 7-day rolling context builder
│   │   ├── health_service    #   HealthKit / Health Connect (HR, resting HR, HRV)
│   │   ├── ble_heart_rate_service  # BLE 0x180D scan/connect/stream + RR parsing + BleHrSession/BleHrvMetrics
│   │   ├── ble_source_provider  # ★ Extensible BLE source plugin system (abstract + registry + service)
│   │   ├── sources/          #   Community signal source implementations
│   │   │   ├── ads1299_source  # ADS1299 8-Ch EEG (EAREEG boards)
│   │   │   └── source_registry_init  # One-line registration entry point
│   │   ├── location_service  #   Geolocator wrapper
│   │   ├── ambient_scan_service  # Environment API
│   │   ├── gps_metrics_service   # Real-time GPS metrics
│   │   ├── calendar_service  #   Device calendar read
│   │   ├── notification_service  # Local notification scheduling
│   │   └── permission_service    # Permission orchestration
│   ├── widgets/              # Shared low-level widgets
│   │   ├── live_hr_waveform  #   60 fps ECG-style PQRST waveform (BLE HR)
│   │   ├── live_signal_chart #   Multi-channel real-time signal chart (any source)
│   │   ├── spectral_analysis_chart  # 3-view FFT spectrum / waterfall / band power
│   │   ├── bci_decoding_view #   Demo neural state classifier
│   │   └── bci_monitoring_view  # Demo signal quality dashboard
│   └── theme/                # Material 3 (light + dark)
├── features/
│   ├── onboarding/           # Step-by-step permission flow
│   ├── journal/              # Journal tab — paginated daily narrative
│   ├── body_blog/            # Blog screen & widgets (detail, cards, AI badge)
│   ├── patterns/             # AI-derived trends & insights
│   ├── capture/              # Manual capture with data-source toggles
│   ├── sources/              # Signal source browser + live signal monitor
│   ├── shell/                # AppShell (3-tab nav) + DebugScreen
│   ├── environment/          # Detailed environment view
│   ├── shared/               # Reusable widgets
│   ├── ai_settings/          # Bring-your-own-AI provider settings
│   └── …                     # health, location, calendar, ai_test, permissions
└── main.dart                 # App entry point

Key Domain Types

Type	Purpose
BodyBlogEntry	Immutable day record: headline, summary, full text, mood, tags, user note, aiGenerated flag, raw BodySnapshot.
BodySnapshot	Flat struct of every collected metric for one day: steps, calories, distance, sleep, avgHeartRate, restingHeartRate, hrv, workouts, environmental data, calendar events. Serialisation-ready for persistence and AI prompts.
BleHrReading	Single BLE HR measurement: bpm + rrMs list (RR intervals in ms for real-time HRV). Emitted by BleHeartRateService.
CaptureEntry	Point-in-time snapshot (health + env + location + calendar + nutrition + signal session). Carries isProcessed / processedAt for the refresh pipeline, plus optional CaptureAiMetadata.
CaptureAiMetadata	AI-derived per-capture metadata: summary, themes, energy level, mood assessment, tags, notable signals, nutritionContext. Stored as JSON in the captures table.
SignalSession	Recorded multi-channel data from any BleSourceProvider: source id/name, channel descriptors, sample rate, timestamped samples. Persisted as JSON blob in CaptureEntry.
NutritionLog	Scanned food product: barcode, product name, brand, Nutri-Score, NOVA group, NutritionFacts per 100 g / per serving. Stored inline on CaptureEntry and in the nutrition_logs table.
JournalAiResult	Parsed AI output: headline, summary, full body, mood, mood emoji, tags.

Provider Architecture

All services are exposed as Provider<T> singletons via service_providers.dart. Screens consume them with ref.read(someServiceProvider), which guarantees a single SQLite connection, makes every service replaceable in tests, and eliminates scattered state.

---

How the AI Journal Works

Smart Refresh Pipeline

getTodayEntry()                       ← called on every Journal visit
  │
  ├─ INSTANT      — persisted entry exists, 0 unprocessed captures
  │                  → return immediately (no sensors, no AI, no network)
  │
  ├─ INCREMENTAL  — persisted entry exists, N unprocessed captures
  │                  → AI runs on new captures only → persist → mark processed
  │
  └─ COLD START   — no entry for today
                    → collect sensors → local template → AI enrich → persist

Capture → Journal Flow

1. CaptureService (manual) or BackgroundCaptureService (WorkManager) creates a capture.
2. Each starts with is_processed = 0.
3. getTodayEntry() detects unprocessed captures → runs AI only when needed.
4. On success, markCapturesProcessed() flips the flag.
5. Next visit → zero unprocessed → instant return from cache.

AI Details

• Endpoint: Default POST ai.governor-hq.com, or any user-configured provider (see Bring Your Own AI below).
• System prompt: First-person voice — "the body speaking to its person." Warm, poetic, data-grounded. Never clinical, never medical advice.
• Context window: 7-day rolling history fed into every prompt (ContextWindowService).
• Fallback: On timeout (45 s) or parse error, the local template entry is returned unchanged.
• aiGenerated flag: Persisted in SQLite so the ✦ badge survives restarts.
• regenerateWithAi(date): Force a fresh AI pass for any date from the journal detail screen.

Mood Inference (v1 — Heuristic)

sleep ≥ 7 h  AND  steps ≥ 5 000  AND  HR > 0   →  energised
sleep < 5 h                                       →  tired
steps ≥ 8 000                                     →  active
AQI > 100                                         →  cautious
sleep ≥ 7 h                                       →  rested
steps = 0  AND  calories = 0                      →  quiet
default                                           →  calm

Will be replaced by a classifier or LLM prompt once sufficient labelled data exists.

Bring Your Own AI

By default Miruns uses its built-in cloud gateway (ai.governor-hq.com) — no API key needed. Users can switch to any OpenAI-compatible provider from More → AI Services:

Provider	Base URL	Notes
Miruns Cloud (default)	ai.governor-hq.com	Built-in, zero config
OpenAI	api.openai.com	GPT-4o, GPT-4o-mini, o1, o3-mini
OpenRouter ★	openrouter.ai/api	300+ models behind one key — OpenAI, Anthropic, Google, Meta, Mistral …
Groq	api.groq.com/openai	Ultra-fast Llama & Mixtral inference
Mistral AI	api.mistral.ai	Mistral Small / Medium / Large
DeepSeek	api.deepseek.com	DeepSeek V3 & R1 reasoning
Together AI	api.together.xyz	Open-source models at scale
Fireworks AI	api.fireworks.ai/inference	Fast open-model inference
Perplexity	api.perplexity.ai	Search-augmented AI
Local (Ollama / LM Studio)	localhost:11434/v1	Private, on-device inference
Custom	user-defined	Any OpenAI-compatible endpoint

All providers speak the same OpenAI chat completions protocol, so no adapter libraries are needed. OpenRouter is recommended when you want a single API key for every major model.

• API keys are stored locally in SQLite — never sent to Miruns Cloud.
• Switching providers takes effect immediately — all AI services (JournalAiService, CaptureMetadataService, etc.) automatically route through the new endpoint via Riverpod's reactive graph.
• A built-in Test Connection button validates the config before saving.

---

Screens

#	Screen	Route	Description
1	Onboarding	/onboarding	Permission flow — location, health, calendar. Every step skippable.
2	Journal	/journal (Tab 0)	Paginated daily blog. Swipe between days. Pull-to-refresh + "Refresh day" on today's card.
3	Journal Detail	—	Full narrative: Sleep, Movement, Heart, Environment, Agenda. AI regeneration & mood/note editor.
4	Patterns	/patterns (Tab 1)	Energy distribution, top themes, keyword tags, notable signals, recent moments timeline.
5	Capture	/capture (Tab 2)	Manual capture with toggleable data sources. BLE HR chip opens a device scanner; once connected, a live ECG-style waveform slides in. Scan Food chip opens a barcode scanner; scanned products show Nutri-Score, macros, and feed into the AI prompt for nutrition-health correlation.
6	Environment	/environment	Expanded environmental data view.
7	AI Services	/ai-settings	Choose AI provider, enter API key, set model, test connection. Supports 11 providers including local inference.
8	Debug	/debug	Raw sensor readouts — health metrics, GPS, ambient data, calendar events.
9	Source Browser	/sources	Browse all registered BLE signal sources (ADS1299, community boards). Each card shows channel count, sample rate, and hardware name.
10	Live Signal	/sources/:sourceId	Full-screen multi-channel signal monitor: scan → pick device → connect → stream. Four view modes via popup menu — Waveform (time-domain), Spectral (FFT spectrum/waterfall/bands), Decoding (demo neural state classifier), Monitor (demo signal quality dashboard). Channel toggle chips, solo mode, recording.

---

Development

Day-to-day

flutter analyze          # static analysis — must pass with zero errors
flutter test             # run the test suite
dart format lib/         # format everything
flutter test --coverage  # generate coverage/lcov.info

Hot reload: r · Hot restart: R · Quit: q

Journal Pipeline Layers

Understanding these three layers prevents accidental slowdowns:

Layer	File	Responsibility
Persistence	local_db_service.dart	SQLite CRUD for entries, captures, settings
Orchestration	body_blog_service.dart	Smart refresh, sensor collection, AI enrichment
AI	journal_ai_service.dart	Prompt building, API call, JSON parsing

Key rule: getTodayEntry() is called on every Journal visit. The fast path (persisted entry + zero unprocessed captures) must return with no I/O beyond two DB queries.

Adding a New Service

1. Create lib/core/services/your_service.dart.
2. Expose a Riverpod Provider<YourService> in service_providers.dart.
3. Wire into BodyBlogService or the relevant feature screen.

Adding a New Feature Screen

1. Create lib/features/your_feature/screens/your_screen.dart.
2. Add a route in lib/core/router/app_router.dart.
3. For a top-level tab, add a StatefulShellBranch to the shell route.

Database Migrations

Schema version lives in local_db_service.dart (_schemaVersion, currently 11).

1. Bump _schemaVersion.
2. Add an if (oldVersion < N) block in _onUpgrade().
3. Wrap ALTER TABLE statements in try/catch for duplicate-column safety (hot-restart resilience).

Testing on Emulators

Data source	Emulator / test setup
Health (Android)	Install Health Connect, add test data manually
Health (iOS)	Physical device required — HealthKit unavailable in Simulator
Location	Set coordinates via emulator extended controls
Calendar	Add events in the device calendar app
Environment	Requires valid GPS to query ambient-scan API
BLE HR	Physical BLE device required — Bluetooth is unavailable in Android Emulator & iOS Simulator. Use a Polar H10, Wahoo TICKR, or any 0x180D device.

All data fetches have 5–15 s timeouts. The app shows zero/empty states for unavailable sensors — never fake data (see CODING_PRINCIPLES.md).

Pre-commit Checklist

• flutter pub get — no unresolved deps
• dart format lib/ — formatter applied
• flutter analyze — 0 errors, 0 warnings
• flutter test — all green
• New screens wired in app_router.dart
• README architecture tree & screen list updated
• Roadmap checkbox ticked if feature was shipped
• No fake data

---

Principles

• No fake data. Sensor unavailable → zero or null, never a plausible placeholder.
• No location tracking. GPS read once for the environment fetch, then discarded.
• Read-only health access. Never writes to HealthKit or Health Connect.
• Graceful degradation. Every data fetch has a timeout. The app never freezes waiting for a sensor.

These are enforced by CODING_PRINCIPLES.md.

---

Roadmap

Shipped

• LLM-backed narrative generation (captures-first, snapshot fallback, local template on failure)
• SQLite persistence for entries, captures, and settings (schema v7)
• 7-day rolling context window for AI prompts
• User annotations (free-text note + mood emoji per day)
• Smart refresh — instant cache, incremental AI, capture-processed tracking
• Background captures via WorkManager (quiet hours, battery awareness)
• Manual capture with configurable data-source toggles
• "Refresh day" button for user-triggered full sensor + AI refresh
• Per-capture AI metadata extraction (CaptureMetadataService)
• Patterns screen — progressive AI-derived insights from capture history
• Share journal entries (share_plus)
• CI pipeline with APK artefact upload
• Resting heart rate + HRV (SDNN) from HealthKit / Health Connect — included in BodySnapshot and AI prompts
• BLE Heart Rate Profile (0x180D) — device scan, connect, live BPM stream with RR intervals
• Live ECG-style PQRST waveform widget (LiveHrWaveform) in the Capture screen
• Continuous BLE HR session recording — timestamped BPM samples accumulated for the full connection duration
• Real-time HRV from BLE RR intervals — RMSSD, SDNN, mean-RR computed at capture time (BleHrvMetrics)
• Full BleHrSession (samples + RR + HRV) stored per capture in SQLite (schema v9)
• BLE HR narrative fed to AI — BPM arc + autonomic tone hint (hrvContext, hrArc) in CaptureAiMetadata
• Bring Your Own AI — plug any OpenAI-compatible provider (OpenAI, OpenRouter, Groq, Mistral, DeepSeek, Together, Fireworks, Perplexity, Ollama, custom)
• Barcode nutrition scanner — scan food products via Open Food Facts API, log Nutri-Score / NOVA / macros per capture
• AI nutrition-health correlation — nutritionContext in CaptureAiMetadata links sugar intake and ultra-processing to next-day HRV / energy patterns
• nutrition_logs table (schema v10) for longitudinal food-tracking queries
• Extensible BLE signal source plugin system — abstract BleSourceProvider, BleSourceRegistry, generic scan/connect/stream engine
• ADS1299 8-channel EEG source (EAREEG boards) — first community source implementation
• Multi-channel real-time signal chart (LiveSignalChart) — per-channel autoscale, toggle/solo, glow aesthetic
• Source Browser + Live Signal screens (/sources, /sources/:sourceId)
• SignalSession model with JSON persistence in CaptureEntry (schema v11)
• Community source contribution guide — CONTRIBUTING_SOURCES.md
• Pure-Dart FFT engine (FftEngine) — Cooley-Tukey radix-2, Hanning window, PSD, EEG band extraction (δ/θ/α/β/γ)
• Spectral analysis chart — live frequency spectrum, waterfall spectrogram, animated EEG band power meters
• BCI decoding demo — 5-state neural classifier (Focus/Relax/Motor-L/Motor-R/Meditate) with confidence ring and timeline
• BCI monitoring demo — per-channel SNR, impedance, artifact detection, data-readiness gauge
• 4-mode visualisation system (SignalViewMode) with animated popup menu switcher

Next — BLE Peripherals

• Continuous background BLE data collection while app is backgrounded
• BLE pulse oximeters (SpO₂) and smart scales
• More community signal sources (OpenBCI Cyton, Muse S, Ganglion, …)

Next — Home Automation

• Smart-home integration (Matter, HomeKit, MQTT)
• Physiological-state-driven rules (sleep → dim lights, wake → start coffee)
• User-defined trigger engine based on body + environment state

Future

• On-device ML for mood/state classification (upgrade BCI decoding from demo heuristics to real inference)
• Multi-user household awareness (BLE presence + individual profiles)
• Wearable companion (Wear OS / watchOS)
• Export to structured health records (FHIR-compatible)

---

License

MIT