PeakPulse HealthTech

A gamified wellness platform that harnesses Hybrid Quantum Machine Learning to predict user churn, personalize fitness challenges, and drive sustainable, long-term health improvements.

Comment

Inspiration

We noticed a recurring pattern in the health and wellness space: millions of people start fitness journeys, but the vast majority drop out within the first three months. While existing apps track metrics, they act as passive observers. We asked ourselves: What if an app could actively predict when a user is losing motivation and intervene before they quit?

This inspired us to build QuantumS, a platform that leverages cutting-edge Hybrid Quantum Machine Learning alongside behavioral gamification to ensure users stay on track.

What it does

QuantumS is a gamified wellness platform that tracks daily activities (Calories, Water, Vitals) directly and via Google Fit integration.

However, its standout feature is the Quantum ML Analytics Engine. It continuously analyzes a user's engagement metrics to predict their churn/dropout risk. When the system detects a high risk of dropping out, it automatically intervenes by offering personalized challenges, unlocking special badges, and sending motivational chatbot nudges to re-engage the user.

How we built it

We divided the architecture into a modern web stack and an advanced mathematics/ML microservice:

  1. Frontend & Core Backend: Built using Next.js 16 (App Router), React 19, and Tailwind CSS. We used NextAuth for secure authentication and MongoDB to store user profiles, stats, and gamification state.
  2. Health Data Integration: We implemented OAuth 2.0 to securely connect with the Google Fit REST API, pulling real-time activity and vitals data.
  3. Hybrid Quantum ML Service: We wrote a standalone FastAPI (Python) microservice. It utilizes a Classical Ensemble ML model (Voting Classifier) fused with a 4-Qubit Variational Quantum Circuit (VQC).

We calculated the final dropout probability prediction utilizing a weighted hybrid fusion approach:

$$ P_{final} = \alpha P_{classical} + \beta P_{quantum} $$

Where the fusion weights \(\alpha\) and \(\beta\) dynamically adapt based on model confidence and feature complexity, typically starting at an equilibrium of \(\alpha = 0.5, \beta = 0.5\).

Challenges we ran into

  • Quantum Simulation Overhead: Training the 4-qubit parameterized quantum circuit with Adam optimizer on classical hardware was computationally heavy. We had to heavily optimize our data normalization pipelines to make the training converge within a reasonable timeframe.
  • API Microservice Communication: Bridging the Next.js TypeScript environment with the Python FastAPI ML environment synchronously without blocking UI rendering required careful asynchronous state management and caching.
  • Google Fit OAuth: Managing access tokens, refresh tokens, and granular permission scopes for Google Fit in a modern Next.js App Router environment posed a steep learning curve.

Accomplishments that we're proud of

  • We successfully developed a functional Variational Quantum-Classical machine learning pipeline moving beyond buzzwords to implement actual parameterized quantum gates bridging with classical ML ensembles. Our classical model hit 93.7% accuracy, and our fused hybrid model successfully isolated complex, non-linear correlations.
  • Creating a beautiful, highly responsive gamification UI that makes tracking health actually feel fun and rewarding.

What we learned

  • Quantum Machine Learning (QML): We learned the deep mathematics behind quantum state superposition, entanglement, and how to map continuous classical data into quantum states using rotation gates like \( R_x, R_y, R_z \).
  • Advanced React Patterns: Architecting a complex dashboard using Next.js Server Components mixed with Client Components for charting (via Recharts).

What's next for QuantumS

  • Real Quantum Hardware: Transitioning our simulated 4-qubit VQC to run on actual IBM Quantum hardware via QiskitRuntimeService.
  • Dietary Computer Vision: Implementing a feature to scan meals via the smartphone camera to automatically log macro-nutrients.
  • Expanded Social Leagues: Building out team-based leaderboards where friends can pool their activity data to compete in global wellness leagues.

Built With

Share this project:

Updates