NeuroAlert

Inspiration

We were inspired by a powerful idea: what if the brain could act as an early-warning system for illness — even before any symptoms appear? People often say, “I felt something was wrong,” long before getting a diagnosis. With emerging neuroscience research showing subtle brainwave changes before clinical symptoms, we saw an opportunity. That’s how NeuroAlert was born — a tool that listens to your brainwaves using AI and EEG signals to detect possible early health risks.

What it does

NeuroAlert analyzes simulated real-time EEG data and classifies the user's cognitive state into three categories:

🟢 Normal
🟡 Mild Cognitive Disturbance
🔴 Alert (Requires User Attention) Each prediction includes a confidence score and a simple, human-readable explanation. The goal is to help users become proactive about their health using the intelligence already within their brains.

How we built it

Frontend: Built with Next.js and TailwindCSS, with multilingual support via i18next for Arabic and English.
Backend: Developed with FastAPI (Python) serving RESTful endpoints.
Database: MongoDB for storing user sessions and predictions.
AI Model: CNN built with TensorFlow/Keras, trained on open EEG datasets from Kaggle.
Visualization: Live EEG-style waveform using Chart.js.
Simulation: Custom Python-based EEG signal generator to mimic real-world input patterns.

Challenges we ran into

Finding EEG datasets with clinical labeling was difficult. Most available datasets were emotion- or task-based, not diagnostic.
EEG signals are inherently noisy and user-specific, which made training robust models a challenge.
Designing a responsive bilingual UI with RTL Arabic support required custom layout logic.
Balancing functionality with responsibility: We were careful not to trigger health anxiety or overstate the tool’s predictions.
Keeping the AI model modular and ready for future integration with real EEG hardware like Muse or OpenBCI.

Accomplishments that we're proud of

Developed a fully functional full-stack simulation of an EEG analysis system.
Trained a deep learning model that can classify brainwave states with high confidence.
Created a bilingual, accessible interface that supports both Arabic and English.
Successfully visualized real-time EEG signal simulation with responsive updates.
Maintained a strong focus on ethical AI design throughout the process.

What we learned

EEG data is fascinating but complex — understanding the brain requires both data science and neuroscience literacy.
Ethical design matters in health-related systems, especially when providing predictions that users may act upon.
Supporting multilingual, RTL-friendly user interfaces added great value but required special handling and testing.
Training temporal models (CNN/LSTM) on non-standardized datasets requires creative preprocessing and augmentation techniques.

What's next for NeuroAlert

Hardware Integration: Connect with real EEG devices such as Muse or OpenBCI for live brainwave input.
Health Timeline: Allow users to track their cognitive state over time and detect trends.
Medical Collaboration: Partner with neuroscientists and doctors to better label and validate brainwave patterns.
Expanded AI Capabilities: Extend detection to conditions like sleep apnea, early dementia, or attention disorders.

NeuroAlert is not just a tool — it’s a mission to turn the brain into our first ally in health, not our last.