LifeGuard: Wearable Health & Environmental Monitoring System

Accessible health and environmental monitoring for everyone, everywhere.

View Demo · Report Bug · Request Feature

Overview

LifeGuard is an innovative wearable system that bridges critical gaps in personal safety, accessibility, and preventive healthcare. By integrating advanced sensors with machine learning algorithms, it delivers real-time insights on health metrics and environmental parameters, making safety monitoring accessible and affordable for all.

The Problem We Solve

Current health and environmental monitoring systems face several critical limitations:

Fragmented Solutions: Most market solutions require multiple devices for comprehensive monitoring, leading to higher costs and added complexity
Limited Accessibility: Premium devices ($400-600) exclude vulnerable populations who need them most
Delayed Response: Many existing solutions fail to provide real-time alerts and updates, limiting their ability to respond promptly to critical situations
Missing Integration: Health and environmental data remain siloed, preventing holistic risk assessment

Our Solution

LifeGuard, powered by the advanced Arduino Nicla Sense ME board, integrates 9 sensors to deliver seamless real-time monitoring of health metrics and environmental conditions at 60% lower cost than premium alternatives like Apple Watch.

60%
Cheaper than Apple Watch

72h
Battery Life

9
Integrated Sensors

IP67
Water Resistance

99.5%
Fall Detection Accuracy

Technical Specifications

Hardware Components

Core Board	Arduino Nicla Sense ME with 9 integrated sensors
Processor	32-bit Cortex-M4 microcontroller at 64MHz
Health Sensor	MAX30102 - Heart Rate & Pulse Oximeter Module
Power	3.7V LiPo Battery (400mAh)
Battery Life	72 hours with optimized power management
Power Consumption	<10mA average draw with dynamic sensor sampling
Durability	IP67 water and dust resistance
Weight	Approximately 45g (including enclosure)
Connectivity	BLE 5.0, WiFi (via companion device)
Display	LCD Screen for local data visualization
Built-in Sensors	• 6-Axis IMU (Accelerometer & Gyroscope) • Temperature & Humidity sensors • Barometric pressure sensor (high-linearity, high-accuracy) • Magnetometer • Gas sensors (VOCs, VSCs, CO, H₂) with AI processing • Sensor fusion for absolute spatial orientation

Software Architecture

Frontend	Backend	Mobile	ML & Analytics
• React 18 • TypeScript • Tailwind CSS • MapBox API	• .NET 8.0 • Node.js • PostgreSQL • Firebase (Real-time DB) • JWT Auth • OAuth 2.0	• Flutter 3.19 • Provider State • Material 3 Design • SharedPreferences • Dark/Light Themes	• LSTM Networks • TinyML Models • Edge Impulse Platform • Edge Inference • Time-series Analysis • Sensor Fusion • Z-score Normalization • Quantization & Pruning

Hosting & Infrastructure

Frontend: Vercel (Web hosting with global CDN)
Backend: Render (API hosting with auto-scaling)
Database: Neon (PostgreSQL hosting)
Real-time Database: Firebase
CI/CD: Automated deployment pipelines

Machine Learning & Activity Recognition

Edge Impulse Integration

LifeGuard incorporates state-of-the-art machine learning capabilities through Edge Impulse for real-time activity classification and fall detection directly on the device:

View Live Edge Impulse Project

Model Specifications

Model Type: Accelerometer-based activity classification with LSTM architecture
Target Device: Arduino Nicla Vision (Cortex-M7 480MHz) / Nicla Sense ME (Cortex-M4 64MHz)
Training Data: 284 samples across 18+ minutes (collected: 18m 5s)
Performance Metrics:
- Validation Accuracy: 100.0%
- Test Accuracy: 99.5%
- Latency: 2ms (real-time capable)
- False Positive Rate: <0.5%
Memory Footprint:
- RAM Usage: 1.8K
- Flash Usage: 17.0K
Optimization: Quantized (int8) for efficient embedded deployment

Activity Classifications

Walking: Normal walking activity detection with gait analysis
Still: Stationary/resting state recognition for baseline monitoring
Falling: Critical fall event detection with 99.5% accuracy
Unknown: Unclassified movement patterns flagged for review

Data Processing Pipeline

Input Configuration:

Sensors: AccX, AccY, AccZ (3-axis accelerometer)
Sampling Rate: 10Hz for optimal battery/accuracy balance
Window Size: 1-second data windows (1000ms)
Window Increase: Sliding window approach for continuous monitoring

Feature Extraction:

Spectral analysis of acceleration patterns
Time-series windowing for motion data
Z-score normalization for sensor data
Sensor fusion combining IMU data

Model Training:

Pre-trained LSTM models for temporal pattern recognition
Transfer learning from established activity datasets
Custom training on device-specific movement patterns
Continuous learning capability for personalization

Real-World Capabilities

This ML model enables:

Automatic Fall Detection: Instant emergency contact notifications upon fall detection
Activity Pattern Analysis: Long-term health insights from movement behaviors
Risk Assessment: Predictive analytics for fall risk based on movement patterns
False Positive Reduction: Correlation with heart rate variability to distinguish falls from jumps
Real-time Processing: On-device inference with <500ms end-to-end latency
Power Efficiency: Optimized model allowing 72h battery life with continuous monitoring

Dataset Overview

Dataset Statistics:

Total Data Collected: 18m 5s
Train/Test Split: 67% / 33%
Sensors Used: accX, accY, accZ @ 10Hz
Labels: falling, still, unknown, walking
Sample Length: 1 second windows

System Architecture

High-Level Architecture

The LifeGuard system follows a distributed architecture with edge computing capabilities:

Edge Layer (Wearable Device):
- Arduino Nicla Sense ME with integrated sensors
- On-device ML inference for real-time fall detection
- Local data preprocessing and filtering
- BLE communication with companion device
Gateway Layer (Mobile/Web):
- Data aggregation from wearable device
- User interface for monitoring and control
- Local caching for offline functionality
- Alert management and notification
Cloud Layer (Backend Services):
- .NET API for data ingestion and processing
- PostgreSQL database with HIPAA-compliant encryption
- Firebase for real-time data synchronization
- Analytics and long-term trend analysis
- Emergency contact management
Integration Layer:
- MapBox for pollution mapping
- MyHealthfinder API for health tips
- Freesound API for wellness sounds
- SendGrid for email notifications
- OAuth providers for authentication

Data Flow Diagram

Data Flow Process:

Data Collection:
- Sensors gather health and environmental data at optimized intervals
- MAX30102 monitors heart rate and SpO2
- Built-in sensors track motion, air quality, and environmental conditions
Edge Processing:
- TinyML models analyze patterns on-device
- Real-time activity classification
- Critical events trigger immediate local alerts
- Data compression before transmission
Data Transmission:
- BLE connection to companion device (smartphone)
- Secure encrypted data packets
- Efficient batching to minimize power consumption
- Automatic reconnection handling
Cloud Processing:
- Data ingestion through REST APIs
- Storage in PostgreSQL with encryption
- Real-time updates via Firebase
- Advanced analytics and pattern detection
User Interface:
- Real-time dashboard visualization on web and mobile
- Interactive pollution maps
- Historical trend analysis
- Customizable alert configurations
Alert System:
- Threshold-based automatic triggers
- Multi-channel notifications (SMS, email, push)
- Emergency contact cascade
- Location sharing with emergency responders

Pictorial System Overview

This diagram illustrates the complete ecosystem showing how the wearable device communicates with various stakeholders:

Wearable User: Direct monitoring and alerts
Healthcare Professional: Access to patient data and trends
Researcher: Anonymous aggregated data for studies
Immediate Family: Emergency notifications and status updates

Hardware Implementation

Physical Hardware Assembly

Components:

Arduino Nicla Sense ME (main processing unit with 9 sensors)
MAX30102 Sensor (heart rate and SpO2 monitoring)
LiPo Battery 3.7V 400mAh (power supply)
Connection wiring and interfaces

System Block Diagram

Arduino Nicla Sense ME Features:

Microcontroller: 32-bit Cortex-M4 @ 64MHz
Smart Sensor Hub: BME688 with AI for gas sensing
IMU: 6-axis motion tracking (BHI260AP)
Pressure Sensor: BMP390 high-accuracy barometric sensor
Connectivity: Bluetooth Module (ANNA-B112) for BLE 5.0
Memory: 2 MB Flash, UART/SPI/I2C interfaces
Power Management: BQ25120A with battery charging
LED Driver: IS31FL3194 for RGB LED control

Pin Configuration

Key Pin Connections:

Power Pins: VIN, 3.3V, GND for power distribution
I2C Interface: SCL, SDA for MAX30102 sensor communication
Analog Pins: A0-A4 for sensor expansion
Digital Pins: D0-D13 for control signals
Battery Connector: JST connector for LiPo battery
USB-C: Programming and charging interface

Hardware Schematics

Schematic Components:

ESLOV Connector: For future expansion and modularity
Battery Connector: JST 2-pin for LiPo battery connection
USB Connector: USB-C for programming, debugging, and charging
Power Management: Voltage regulation and battery charging circuit
Sensor Interfaces: I2C bus connections for external sensors
LED Control: RGB LED driver circuitry

System Wiring Diagram

Connection Details:

Nicla Sense ME to MAX30102: I2C connection (SCL, SDA, VIN, GND)
Battery to Nicla: Direct connection via JST connector
Power Distribution: 3.7V from battery regulated to 3.3V for sensors

Device Enclosure Design

Enclosure Features (Designed in SolidWorks):

Compact wearable form factor
Watch-style wrist mounting system
IP67-rated water and dust resistance
Ventilation for environmental sensors
Secure compartments for electronics
Easy battery replacement design
Integrated watch strap mounting points

Final Device Design

Completed Device:

White protective housing with LED indicator window
Standard watch strap for comfortable wearing
Compact 45g total weight
Dimensions optimized for all-day wear
LCD screen for local display (optional)
Button-free operation (controlled via app)

Getting Started

Prerequisites

Software Requirements:

Node.js 18+ (for web development)
.NET SDK 8.0 (for backend API)
Flutter SDK 3.19+ (for mobile app)
PostgreSQL 15+ (database)
Arduino IDE or PlatformIO (for firmware development)
Git (version control)

Hardware Requirements (for development):

Arduino Nicla Sense ME board
MAX30102 sensor module
LiPo battery (3.7V, 400mAh)
USB-C cable for programming
Computer with Bluetooth capability

Quick Start Guide

1. Clone the Repository

git clone https://github.com/evansachie/LifeGuard.git
cd LifeGuard

2. Set Up Environment Files

# Backend (.NET)
cp backend/.env.example backend/.env

# Node Server
cp node-server/.env.example node-server/.env

# Web Dashboard
cp web/.env.example web/.env

# Mobile App
cp mobile/.env.example mobile/.env

Edit each .env file with your configuration:

Database connection strings
API keys (MapBox, SendGrid, Freesound)
Firebase credentials
OAuth client IDs
JWT secret keys

3. Database Setup

# Install PostgreSQL if not already installed
# Create database
createdb lifeguard_db

# Run migrations (from backend directory)
cd backend
dotnet ef database update

4. Start the Backend Server (.NET)

cd backend
dotnet restore
dotnet build
dotnet run

The API will be available at https://localhost:5001 (or configured port)

5. Start the Node Server

cd node-server
npm install
npm start

The Node server will run at http://localhost:3000 (or configured port)

6. Launch the Web Dashboard

cd web
npm install
npm run dev

Access the dashboard at http://localhost:3000

7. Run the Mobile Application

cd mobile
flutter pub get
flutter run

Select your target device (iOS simulator, Android emulator, or physical device)

8. Flash Firmware to Device

cd firmware
# Using Arduino IDE: Open sketch and upload
# Or using PlatformIO:
pio run --target upload

Project Structure

lifeguard/
├── .github/                    # GitHub actions and templates
├── firmware/                   # Arduino code and sketches
│   ├── test-sketches/          # Sketches to test Nicla Sense ME
│   └── we-dashboard/           # Web dashboard for reading sensor data
├── web/                        # React dashboard
│   ├── public/                 # Static assets
│   ├── src/                    # React components
│   │   ├── components/         # Reusable UI components
│   │   ├── pages/              # Main application views
│   │   ├── services/           # API integrations
│   │   └── store/              # Redux state management
├── mobile/                     # Flutter mobile app
│   ├── lib/                    # Dart code
│   │   ├── models/             # Data models
│   │   ├── screens/            # UI screens
│   │   ├── services/           # Business logic
│   │   └── widgets/            # Reusable components
├── backend/                    # .NET Core API
│   ├── Controllers/            # API endpoints
│   ├── Models/                 # Data structures
│   ├── Services/               # Business logic
│   └── Middleware/             # Request processing
├── node-server/                # Node.js backend service
│   ├── controllers/            # Route controllers
│   ├── models/                 # Database schemas
│   ├── routes/                 # API routes
│   ├── services/               # Business logic
│   ├── middleware/             # Express middleware
│   └── utils/                  # Helper functions
├── docs/                       # Documentation
│   ├── api/                    # API reference
│   ├── images/                 # Project images
│   └── tutorials/              # User guides
└── .devcontainer/              # Development container config

API Documentation

View Complete API Documentation on Postman

The LifeGuard API is split across two backend services for optimal performance and modularity:

.NET Backend Service

Base URL: https://lifeguard-hiij.onrender.com

Handles user authentication, account management, and photo storage.

Authentication Endpoints

Method	Endpoint	Description	Auth Required
GET	`/`	Health check endpoint	No
POST	`/api/Account/login`	User login with email/password	No
POST	`/api/Account/register`	Register new user account	No
POST	`/api/Account/forgot-password`	Initiate password recovery	No
POST	`/api/Account/ResendOTP`	Resend OTP to user	No
POST	`/api/Account/VerifyOTP`	Verify user OTP	No
POST	`/api/Account/ResetPassword`	Reset password with token	No
POST	`/api/Account/CompleteProfile`	Complete user profile setup	Yes
GET	`/api/Account/{id}`	Get account info by ID	Yes
GET	`/api/Account/GetProfile/{id}`	Get detailed profile	Yes
GET	`/api/Account/google-login`	Initiate Google OAuth	No
GET	`/api/Account/signin-google`	Google OAuth callback	No
DELETE	`/api/Account/{id}`	Delete user account	Yes

Photo Management Endpoints

Method	Endpoint	Description	Auth Required
POST	`/{id}/photo`	Upload user photo	Yes
DELETE	`/{id}/photo`	Delete user photo	Yes
GET	`/{id}/photo`	Retrieve user photo	Yes

Node.js Backend Service

Base URL: https://lifeguard-node.onrender.com

Handles health metrics, emergency contacts, medications, and advanced AI features.

Memo Endpoints

Method	Endpoint	Description	Auth Required
GET	`/api/memos`	Get user memos	Yes
POST	`/api/memos`	Create new memo	Yes
GET	`/api/memos/undone/count`	Count of incomplete memos	Yes

Emergency Contact Endpoints

Method	Endpoint	Description	Auth Required
GET	`/api/emergency-contacts`	Get emergency contacts	Yes
POST	`/api/emergency-contacts`	Add emergency contact	Yes
POST	`/api/emergency-contacts/alert`	Send emergency alert	Yes
GET	`/api/emergency-contacts/test-alert/{id}`	Test alert to contact	Yes
GET	`/api/emergency-contacts/verify`	Verify contact with token	No
GET	`/api/emergency-contacts/alerts`	Get alert history	Yes

Health Metrics Endpoints

Method	Endpoint	Description	Auth Required
GET	`/api/health-metrics/latest`	Get latest metrics	Yes
POST	`/api/health-metrics/save`	Save new metrics	Yes
GET	`/api/health-metrics/history`	Get metrics history (last 10)	Yes

Exercise Endpoints

Method	Endpoint	Description	Auth Required
GET	`/api/exercise/stats`	Get exercise stats & streaks	Yes
POST	`/api/exercise/complete`	Record workout session	Yes
POST	`/api/exercise/goals`	Set/update workout goals	Yes
GET	`/api/exercise/workout-history`	Get workout history	Yes
GET	`/api/exercise/calories-history`	Get calories history	Yes
GET	`/api/exercise/streak-history`	Get streak history	Yes

Medication Endpoints

Method	Endpoint	Description	Auth Required
GET	`/api/medications`	Get all medications	Yes
POST	`/api/medications/add`	Add new medication	Yes
PUT	`/api/medications/:id`	Update medication	Yes
DELETE	`/api/medications/:id`	Delete medication	Yes
POST	`/api/medications/track`	Track dose (taken/skipped)	Yes
GET	`/api/medications/compliance`	Get compliance rate	Yes
GET	`/api/medications/emergency/:userId`	Emergency medication info	No

Health Tips Endpoints

Method	Endpoint	Description	Auth Required
GET	`/api/health-tips`	Get health tips from MyHealthfinder	Yes
GET	`/api/health-tips/topic/:id`	Get specific topic details	Yes

Sound & Wellness Endpoints

Method	Endpoint	Description	Auth Required
POST	`/api/freesound/audio-proxy`	Proxy Freesound audio	Yes
GET	`/api/favorite-sounds`	Get all favorite sounds	Yes
GET	`/api/favorite-sounds/{userId}`	Get user favorites	Yes
DELETE	`/api/favorite-sounds/{userId}/{soundId}`	Remove from favorites	Yes

RAG System Endpoints (AI Document Q&A)

Method	Endpoint	Description	Auth Required
POST	`/api/upload`	Upload PDF for processing	Yes
POST	`/api/ask`	Ask question about PDFs	Yes

RAG System Features:

Upload health documents (prescriptions, lab reports, medical records)
Intelligent document parsing and text extraction
Vector embeddings for semantic search
Natural language question answering
Context-aware responses with source citations

Voice Commands Endpoints

Method	Endpoint	Description	Auth Required
POST	`/api/voice-commands/process`	Process voice command with NLP	Yes
POST	`/api/voice-commands/emergency`	Emergency voice command	Yes
GET	`/api/voice-commands/commands`	Get available commands	Yes

Supported Voice Commands:

"Check my heart rate"
"Show air quality"
"Call emergency contact"
"Start meditation"
"Record workout"

User Preferences Endpoints

Method	Endpoint	Description	Auth Required
GET	`/api/user-preferences/notifications`	Get notification settings	Yes
POST	`/api/user-preferences/notifications`	Update notification settings	Yes
POST	`/api/user-preferences/send-test-email`	Send test notification	Yes

Implementation Timeline

Core Developers

_{Evans Acheampong}
Full Stack & Hardware Lead
University of Ghana

Responsibilities:

Hardware integration & sensor optimization
Firmware development (Arduino/C++)
Frontend development (React, Flutter)
Node.js backend services
User interface design & testing
System documentation
Project management

_{Michael Adu-Gyamfi}
Backend & ML Lead
University of Ghana

Responsibilities:

Backend development (.NET, PostgreSQL)
Machine learning model development
Edge Impulse ML pipeline
Firmware optimization
Data analytics & processing
System security (encryption, CI/CD)
Firebase real-time database

Academic Advisors

_{Dr. Percy Okae}
Project Supervisor
Department of Computer Engineering
University of Ghana

Provided academic guidance, technical supervision, and project oversight throughout the development process.

_{Chiratidzo Matowe}
Project Advisor
University of Ghana

Offered technical advice on system architecture, user experience design, and industry best practices.

_{Marvin Rotermund}
Ambassador
Embedded Learning Challenge
Edge Impulse

Provided guidance on machine learning implementation, Edge Impulse platform utilization, and embedded AI optimization.

Screenshots

Web Dashboard

Main dashboard showing real-time health metrics, environmental data, and activity summary

Analytics Panel

Comprehensive analytics with historical trends, charts, and insights

Sticky Notes & Task Management

Memo and task management system for health reminders and daily notes

Health Report

Detailed health report with metrics, trends, and personalized insights

Pollution Tracker Map

Interactive MapBox-powered pollution map showing air quality data and user location

Health Tips

Personalized health tips from MyHealthfinder API based on user profile and health data

Emergency Contacts

Emergency contact management with verification system and test alert functionality

Help & Support

Comprehensive help center with FAQs, tutorials, and support resources

User Profile

User profile management with health information, preferences, and account settings

Exercise Routines

Exercise tracking with workout routines, calories burned, and streak monitoring

Wellness Hub - Breathing Exercises

Guided breathing exercises for stress relief and relaxation

Wellness Hub - Meditation

Meditation sessions with timer and ambient sound options

Wellness Hub - Zen Sounds

Curated ambient sounds from Freesound API for relaxation and focus

Health Metrics Calculator

Interactive calculator for BMI, BMR, ideal weight, and other health metrics

Device Dashboard - Live Sensor Data

Real-time device dashboard showing live sensor data streams with interactive controls

Bill of Materials

Core Components

Item	Description	Cost (GH₵)	Quantity	Total (GH₵)
Arduino Nicla Sense ME	Main processing unit with 9 integrated sensors (IMU, temp, humidity, pressure, magnetometer, gas sensors)	1,500.00	1	1,500.00
MAX30102 Sensor	Heart Rate & Pulse Oximeter Module with I2C interface	64.00	1	64.00
LiPo Battery	3.7V 400mAh rechargeable battery with JST connector	95.00	1	95.00
LCD Screen	Small display for local data visualization (optional)	90.00	1	90.00
Custom Enclosure	3D-printed housing with watch straps	~50.00	1	50.00

Total Estimated Cost: GH₵ 1,799.00 (~$113 USD)

Additional Development Costs (Not per-unit)

Item	Purpose	Cost Range
3D Printer Access	Enclosure prototyping	GH₵ 500 - 1,000
Development Tools	Software licenses, cloud services	GH₵ 1,000 - 2,000
Testing Equipment	Multimeters, oscilloscope access	GH₵ 500 - 1,500
PCB Prototyping	Custom circuit boards (if scaled)	GH₵ 2,000+

Cost Comparison with Market Alternatives

Device	Price Range	LifeGuard Advantage
Apple Watch Series 9	$399 - $799	60% cheaper, similar sensors
Fitbit Sense 2	$249 - $299	62% cheaper, more environmental sensors
Samsung Galaxy Watch 6	$299 - $429	65% cheaper, open-source software
Garmin Venu 3	$449 - $499	72% cheaper, specialized health focus

Scalability & Manufacturing

Current Cost Structure (Prototype):

Hand-assembled units
3D-printed enclosures
Off-the-shelf components
Estimated cost per unit: GH₵ 1,800

Projected Cost at Scale (1,000+ units):

Injection-molded enclosures: -30%
Bulk component purchasing: -20%
Automated assembly: -15%
Projected cost per unit: GH₵ 900 - 1,100

Target Retail Price: GH₵ 1,500 - 2,000 ($95 - $125)

Live System Access

Web Dashboard https://lifeguard-vert.vercel.app Full-featured web application with: Real-time health monitoring Interactive analytics Pollution mapping Wellness features Device management	Mobile App Flutter App (iOS & Android) Download and install: BLE device pairing Real-time notifications Offline data sync Emergency SOS Activity tracking
.NET API https://lifeguard-hiij.onrender.com/api Core backend services: User authentication Profile management Photo storage OAuth integration	Node.js API https://lifeguard-node.onrender.com Specialized microservices: Health metrics Emergency alerts Medication tracking AI features (RAG, Voice)

Contributing

We welcome contributions from the community! LifeGuard is an open-source project aimed at making health monitoring accessible to all.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Citation

If you use LifeGuard in your research or project, please cite:

@misc{lifeguard2025,
  title={LifeGuard: Wearable Health and Environmental Monitoring System},
  author={Acheampong, Evans and Adu-Gyamfi, Michael Kwabena},
  year={2025},
  institution={University of Ghana},
  url={https://github.com/evansachie/LifeGuard}
}

Acknowledgments

We would like to express our gratitude to:

University of Ghana for providing facilities and academic support
Dr. Percy Okae for invaluable guidance and supervision
Chiratidzo Matowe for technical advice and mentorship
Marvin Rotermund and Edge Impulse for ML platform and support
Arduino for the amazing Nicla Sense ME platform
Our user testers for valuable feedback and patience
Open-source community for tools and libraries that made this possible

Name		Name	Last commit message	Last commit date
Latest commit History 1,458 Commits
.vscode		.vscode
backend		backend
docs		docs
firmware		firmware
mobile		mobile
node-server		node-server
rag		rag
web		web
.gitattributes		.gitattributes
.gitignore		.gitignore
.hintrc		.hintrc
LICENSE		LICENSE
README.md		README.md