NeuroFlow AI: Predictive Biofeedback for Med

• 

  https://edgehealth.azurewebsites.net/

• 

  Prediction using a smartwatch

• 

EdgeHealth · NeuroFlow AI Immersive Edge AI for Cognitive Wellness

🧩 About the Project EdgeHealth is an XR platform designed to support real‑world health and wellness use cases through immersive spatial experiences. It enables users to interact with real‑time health insights inside an intuitive XR environment, improving engagement, understanding, and decision‑making. The platform is deployed on Microsoft Azure 👉 https://edgehealth.azurewebsites.net ensuring scalable backend infrastructure, reliable APIs, and production‑ready deployment aligned with Microsoft for Startups best practices. By combining XR interaction, Edge AI, and cloud‑backed orchestration, EdgeHealth demonstrates how immersive technology can evolve into a scalable startup solution with real social impact in healthcare and digital wellbeing.

💡 Inspiration Cognitive fatigue and “brain fog” are often the invisible consequences of prolonged physical inactivity. In modern work environments, people spend hours sitting, leading to mental burnout, reduced neuro‑efficiency, and declining focus. The problem is subtle: Most people don’t realize they need a mental reset until it’s already too late. EdgeHealth was inspired by this gap — the absence of real‑time, objective feedback that connects physical behavior to cognitive health.

🧠 The Solution — NeuroFlow AI NeuroFlow AI is an Edge AI system that detects cognitive overload using movement‑based biofeedback. Using wrist‑worn accelerometers, the system monitors movement patterns to infer Cognitive Load, correlating prolonged inactivity with mental fatigue. ✅ How It Works

Motion data is processed locally at the edge A trained AI model classifies user states:

Sitting Standing Stretching

A time‑weighted inactivity filter tracks cognitive stagnation When a critical threshold is reached:

A haptic alert suggests a Cognitive Reset (stand or stretch)

Once movement is detected again:

The cognitive timer resets in real time

All inference happens on‑device — no sensitive data leaves the system.

⚙️ Technical Implementation 📊 Dataset

20,000+ labeled samples Activities: Sitting, Standing, Stretching Optimized for high‑accuracy activity recognition

🧠 Model NeuroWAKEv1

Lightweight architecture optimized for Edge Computing Low‑latency inference Designed to run on constrained environments

Note: For IP reasons, the full XML/BIN models are proprietary. Full access can be provided to judges via a private repository under NDA, if required.

🔒 Privacy‑First Architecture

100% local inference No cloud processing of health data Designed for GDPR‑compliant environments

🧩 Core Logic

Time‑weighted inactivity detection Correlation between physical immobility and mental fatigue Real‑time state transitions validated by sensor feedback

🚧 Challenges We Ran Into The biggest challenge was building a system that is:

Accurate Private Real‑time Non‑intrusive

We had to ensure the AI model could react immediately to user movement without draining power or relying on constant cloud connectivity. Balancing responsiveness with privacy and efficiency required multiple iterations of both model design and signal filtering logic.

🏆 Accomplishments We’re Proud Of

✅ Built a fully local cognitive wellness AI system ✅ Achieved real‑time edge inference with high accuracy ✅ Created an intuitive feedback loop using haptics ✅ Designed a solution that scales from individual users to enterprise wellness programs

Built With

• neurohack