📌 Index

Section	Link
🔍 About	About the Project
✨ Features	Main Features
🐞 Issues	Current Issues
📁 Structure	Folder Structure
🧠 Improvements	Possible Improvements
🗺️ Diagram	Architecture Diagram
⚙️ Usage	Installation and Usage
📊 Performance	Model Performance: XGBoost & Neural Network
🧪 Testing	Testing
🎥 Demo	Demo
🚀 Deployment	Render Deployment
🐳 Docker	Dockerization
🤝 Team	Collaborators

---

🧠 About the Project

This project focuses on predicting the risk of suffering a stroke using a machine learning model trained on a real-world health dataset. The model analyzes key personal, medical, and lifestyle factors to determine whether an individual is at risk. The goal is to support early detection and prevention by providing fast, data-driven assessments through a user-friendly web interface.

---

🔍 Main Features

✅ Complete EDA process with data cleaning and visualizations.
✅ Binary classification using XGBoost Classifier.
✅ Backend built with Python and FastApi.
✅ Frontend developed with Node and React. ✅ Database integration using Supabase.
✅ Modular and scalable project structure.

---

🐞 Current Issues

❌The model's accuracy is limited due to the small dataset size. Increasing the amount and diversity of data could significantly improve performance.

---

💡 Possible Improvements

✅ Add user authentication and data history

---

📁 Folder Structure

# Clasificación_Multiclase_equipo_6
📂 Clasificaci-n_Multiclase_equipo_6
├── 📂 .venv/
├── 📂 .github/
├── 📂 client/
│   └── 📂 public/
│   └── 📂 src/
│       └── 📂 assets/
│       └── 📂 components/
│               └── About.jsx/
│               └── AssesmentForm.jsx/
│               └── Header.jsx/
│               └── Hero.jsx/
│               └── PredictionHistory.jsx/
│       └── App.jsx/
│       └── main.jsx/
├── 📂 data/  
│   └── csv_final.csv
├── 📂 eda/
│   └── EDA.ipynb
│── 📂 model/
│   └── modelo_xgb.pkl
│   └── models.ipynb
│   └── encoded.py              
├── 📂 server/
│   └── 📂 database/
│         └── db_connection.py
│         └── save_user.py
│   └── requirements.txt
│   └── main.py
│   └── dockerfile
├── 📂 tests/
│   └── test_model.py
├── 📜 .env 
├── 📜 README.md  
├── 📜 .gitignore  
├── 📜 requirements.txt
├── 📜 dockerfile
├── 📜 docker-compose.yaml
├── 📜 .dockerignore

---

🧠 Architecture Diagram

---

⚙️ Installation and Usage

1️⃣ Clone the repository

git clone https://github.com/Bootcamp-IA-P4/brainstroke-navm.git
cd brainstroke-navm

2️⃣ Create and activate the virtual environment

python -m venv .venv
source .venv/Scripts/activate # Windows
source .venv/bin/activate  # Linux/Mac

3️⃣ Install backend dependencies and run the API

pip install -r requirements.txt
uvicorn server.main:app --reload

4️⃣ Install and run the frontend

cd client
npm install
npm run dev

---

📊 Model Performance: XGBoost & Neural Network

🔸 XGBoost Classifier

• Type: Binary Classification
• Target: Predict stroke risk based on tabular health data
• Framework: scikit-learn, xgboost
• Performance:
  

---

🔸 Neural Network (MobileNetV2 - Transfer Learning)

• Type: Convolutional Neural Network (CNN)
• Task: Multiclass classification of brain CT images
• Classes: Bleeding, Ischemia, Normal
• Input: 224x224 RGB images
• Base Model: Pretrained MobileNetV2
• Final Layer: nn.Linear(model.last_channel, 3)
• Loss Function: CrossEntropyLoss
• Optimizer: Adam (lr=1e-4)
• Epochs: 6
• Dataset: Custom-labeled CT image dataset
• Model File: model/model2.pth

---

🧐 Testing

Copy the following command to run the tests:

python -m unittest discover

---

📂 Demo

🔗 ▶️ Ver Demo

---

🚀 Deployment on Render

🔹 You can view the live frontend of the project here:
🌐 https://brainstroke-navm-1-front.onrender.com

🔹 You can access the API documentation here:
🛠️ https://brainstroke-navm.onrender.com

ℹ️ Note: It might take a few seconds to load when idle, as free Render services may enter sleep mode.

---

🐋 Dockerization

This application is Dockerized. To run it, execute the following command:

docker-compose up --build

---

🧑‍💻 Collaborators

This project was developed by the following contributors:

• Michael López
• Veida Velázquez
• Nhoeli Salazar
• Alejandro Rajado

---

(⬆️ Back to top)