🧠 NeuroGuard – Mental State AI Assistant

"In 2080, your mind is the battlefield. NeuroGuard is your last line of defence."

---

Demo

Demo.Neuroguard.mp4

💡 Project Name: NeuroGuard

---

🏁 Hackathon Context

NeuroGuard was developed during Great UniHack 2025 (Manchester) — a university hackathon focused on future challenges and speculative problem-solving.

The challenge encouraged teams to:

• identify a realistic problem of the future, and
• design a theoretical or experimental system that could address it.

Our team chose to focus on a challenge we believe will define the coming decades:

Mental health in an era of extreme technological saturation.

---

🛡️ The Solution

NeuroGuard is a lightweight AI-powered web application that estimates a user’s mental state based on everyday lifestyle signals and provides actionable recommendations to prevent overload and burnout.

The system:

• collects structured lifestyle inputs,
• estimates a mental state score (0–100) using a machine learning model,
• generates context-aware suggestions to support cognitive balance.

The application is designed to run locally, with optional AI-enhanced recommendations.

---

🧠 How It Works (High Level)

1. The user enters lifestyle information (sleep, screen time, activity, daylight, etc.).
2. A trained RandomForestRegressor estimates the mental state score.
3. Based on the prediction:
   • offline fallback recommendations are generated, or
   • optional Gemini API suggestions are provided.
4. The result is displayed instantly in a clean, cyberpunk-inspired UI.
5. User history is stored locally for future analysis.

---

🔑 Core Features

• 🧠 Mental State Estimation
  Machine learning–based prediction using a RandomForestRegressor.

• 🧬 AI-Driven Recommendations (Optional)
  Context-aware mental wellness tips via Google Gemini API, with full offline fallback.

• 💻 Interactive Web Interface
  Modern, responsive frontend built with HTML, CSS, and JavaScript.

• 💾 Local Persistence
  User inputs and summaries stored in a local SQLite database.

• 🔐 Privacy by Design
  Runs locally by default. External AI integration is optional and transparent.

• ✅ Robust Validation & UX
  Input validation, range checks, loading states, and session-safe navigation.

---

🧰 Tech Stack

Languages

• Python
• HTML, CSS, JavaScript

Backend

• Flask
• SQLite
• python-dotenv

Machine Learning

• scikit-learn
• pandas
• RandomForestRegressor

AI Integration (Optional)

• Google Gemini API

🛠️ Installation Instructions:

To get started, make sure you have Python 3.9+ installed. Then clone the repository and install all required dependencies using the following command:

pip install -r requirements.txt

After that, create a .env file in the root directory of the project (see .env.example) and add your Gemini API Key in the following format:

FLASK_SECRET_KEY=your-secret-key
GEMINI_API_KEY=your-gemini-api-key   # optional

Then launch the application with:

python main.py

---

🚀 Usage:

1. Visit: http://localhost:4000
2. Submit lifestyle info → Receive mental state score
3. Trigger Gemini-powered suggestions
4. Retrain the model using CLI: python main.py

---

🎯 Why It Matters

NeuroGuard is not a medical device. It is a conceptual exploration of digital well-being in an over-optimised future.

As technology increasingly shapes cognition, projects like NeuroGuard highlight the importance of:

• mental resilience,
• human-centered system design,
• ethical AI integration.

---

"They enhanced our bodies. We built NeuroGuard to protect what’s left of the mind."

👥 Team

• Dmytro Dudarenko
• Kamron Khusainov
• Thomas Palmer

📌 Notes

• This project was developed as a team-based exploratory prototype.
• The ML model is trained on structured sample data for demonstration purposes.
• Designed for educational, experimental, and portfolio use.