Case Study

Reactive Heart — AI-Powered Health Monitoring

A Flutter mobile application that uses facial scanning and computer vision to estimate heart rate, SpO₂, blood pressure, breathing rate, and stress levels — all from a smartphone camera, with no wearables required.

Project Overview

Contactless Vital Signs Monitoring

Reactive Heart represents a paradigm shift in personal health monitoring. Built entirely in-house by StrongMoments Technologies, this mobile application leverages advanced computer vision, machine learning, and real-time signal processing to estimate a user's vital signs through nothing more than a smartphone's front-facing camera.

By analyzing subtle color variations in facial skin caused by blood flow (photoplethysmography via camera), the app extracts heart rate, blood oxygen saturation, blood pressure estimates, breathing rate, and computes a composite stress score — all in under 30 seconds. The backend uses a WebSocket-based pipeline for real-time frame processing and AI-powered ethnicity analysis to improve measurement accuracy across diverse populations.

Product
Reactive Heart
Industry
HealthTech & Wellness
Built By
StrongMoments Technologies
Services
AI, ML, Computer Vision, Mobile
The Challenge

Accessible Health Monitoring Without Hardware

  • Traditional vital signs monitoring requires expensive wearables or clinical equipment, making routine health checks inaccessible to millions
  • Smartphone-based approaches existed but suffered from low accuracy, poor UX, and no real-time feedback during scanning
  • Processing facial video frames for physiological signals demands heavy computation that can't run entirely on-device without draining batteries
  • Accuracy of camera-based PPG varies significantly across different skin tones and ethnicities, requiring calibration
  • Users needed results they could trust, share with doctors, and track over time — not just one-off numbers
Our Solution

A Full-Stack AI Health Platform

  • Built a Flutter mobile app with Google ML Kit face detection for real-time face tracking and validation during scanning
  • Designed a WebSocket-based streaming pipeline that sends camera frames to a FastAPI backend for server-side signal processing
  • Implemented custom ML algorithms for heart rate extraction via PPG, SpO₂ estimation, and blood pressure inference from pulse waveforms
  • Built an ethnicity analysis model that auto-calibrates measurement algorithms based on the user's detected skin tone for improved accuracy
  • Added MedGemma AI integration for medical document scanning and analysis — users can photograph lab reports and get instant AI-powered interpretations
  • Implemented PDF report generation, scan history with charts, step tracking (pedometer), and health reminders
Platform Capabilities

Key Features

face_retouching_natural

Camera-Based Face Scanning

Real-time face detection using Google ML Kit with validation for proper positioning, lighting, and stillness. Animated scan overlay provides live feedback during measurement.

monitor_heart

5 Vital Signs in 30 Seconds

Heart rate (BPM), blood oxygen (SpO₂), blood pressure estimation, breathing rate, and composite stress score — all derived from facial video analysis.

stream

Real-Time WebSocket Pipeline

Camera frames stream to the FastAPI backend via WebSocket for server-side ML processing, returning rolling vitals updates every 5 seconds during scan.

diversity_3

Ethnicity-Aware Calibration

An async ML job analyzes the first captured frame to determine ethnicity, allowing PPG algorithms to auto-calibrate for improved accuracy across diverse skin tones.

clinical_notes

MedGemma AI Document Analysis

Upload photos of lab reports, X-rays, or prescriptions and get structured AI-powered medical analysis with findings, severity assessment, and recommendations.

history

Scan History & Reports

All results are saved with timestamps and trend charts. Users can generate shareable PDF reports and track their health metrics over time.

directions_walk

Step Counter & Activity

Built-in pedometer with GPS-tracked walking routes and daily step goals to encourage an active lifestyle alongside vitals monitoring.

alarm

Health Reminders

Configurable reminders for medication, water intake, and regular health check scans to build consistent health monitoring habits.

translate

Multi-Language Support

Full internationalization with Slang — currently supporting English and Hindi with extensible architecture for additional languages.

Impact

Measurable Results

30s
Full Vitals Scan Time
5
Vital Signs per Scan
0
Wearables Required
24/7
Real-Time Cloud Processing
Architecture

Mobile Client

  • Flutter with BLoC state management for clean, testable architecture
  • Google ML Kit for real-time face detection and face mesh analysis
  • Camera API with custom scan overlay painters and animated feedback
  • WebSocket streaming for frame-by-frame data upload to backend
  • Shorebird for instant OTA updates without app store review cycles
  • Firebase Analytics for user engagement and crash tracking
Architecture

Backend & ML Pipeline

  • FastAPI (Python) with async WebSocket endpoint for real-time processing
  • OpenCV + NumPy for frame decoding and signal processing
  • Custom ML models for PPG extraction, SpO₂, BP, and breathing rate
  • Ethnicity analysis worker with async job queue pattern
  • Gemini AI integration for medical document analysis (MedGemma)
  • PostgreSQL with Alembic migrations, JWT auth, and bcrypt password hashing
Technology

Built With

Flutter Dart BLoC Pattern Google ML Kit FastAPI Python OpenCV NumPy PostgreSQL Alembic WebSocket Gemini AI Shorebird Firebase AWS EC2 JWT Auth
Reactive Heart proves that accessible healthcare technology doesn't have to be complicated. We've put clinical-grade vital signs monitoring into everyone's pocket — no extra hardware needed, just your phone's camera and our AI.
— Engineering Team, StrongMoments Technologies
Explore More

Related Services

Want to build an AI-powered health product?

Let's explore what's possible with computer vision and ML.

Start a Conversation