---

❤️ AI-Powered Heart Disease Prediction Website

📌 Website Demonstration:

📌 Overview

This web-based heart disease prediction application utilizes machine learning models to analyze user-input health data and determine the likelihood of heart disease. The system integrates a neural network (ANN) alongside other classification models to provide an accurate and user-friendly platform for health risk assessment.

Developed using Python, Flask, HTML, CSS, JavaScript, and machine learning models, this application aims to make early detection of cardiovascular disease more accessible and effective.

---

🎯 Key Features

🔍 1. Objective & Dataset

• This model is designed to predict heart disease based on various health factors such as age, cholesterol levels, blood pressure, and heart rate.
• The dataset is a combination of five heart-related datasets, creating one of the most extensive heart disease datasets with 918 unique patient records and 11 key health attributes.

📌 Project Introduction & Dataset Attributes:

---

💻 2. Website User Interface

• Simple and intuitive interface for users to input health data.
• One-click prediction button for instant results.
• Automated Testing Feature for bulk predictions.

📌 Website Interface Screenshot:

---

🧠 3. Machine Learning Models

• Implemented Logistic Regression, Support Vector Machine (SVM), and Artificial Neural Networks (ANN).
• 10-Fold Cross Validation to prevent overfitting and ensure high accuracy.
• Hyperparameter tuning used to optimize the ANN model.
• Feature selection based on correlation analysis.

📌 Model Performance Results:

📌 Classification Reports & Confusion Matrices:

---

⚙️ 4. Data Processing & Training Pipeline

• Data Cleaning & Preprocessing:

  • Handled missing values and outliers.
  • Standardized numerical features.
  • One-hot encoding for categorical variables.

• Training the Models:

  • Used scikit-learn and TensorFlow/Keras for model training.
  • Optimized training using gradient descent and backpropagation.
  • Applied dropout layers to improve ANN generalization.

📌 Data Processing & Training Workflow:

---

🛠️ Installation & Setup Guide

🔑 Prerequisites

Before proceeding, ensure you have the following installed:

• Python 3.x
• Required dependencies in requirements.txt (Flask, NumPy, Pandas, Scikit-Learn, TensorFlow, etc.)

🚀 Setup Instructions

1️⃣ Clone the Repository

git clone https://github.com/your-repo/heart-disease-prediction.git
cd heart-disease-prediction

2️⃣ Install Dependencies

pip install -r requirements.txt

3️⃣ Run the Application

cd Website
python app.py

4️⃣ Access the Website

• Once running, open your browser and navigate to:

  http://127.0.0.1:5000
  

---

⚡ Project Structure

📂 Heart-Risk-ANN-Model_Website
│── 📂 Model
│   │── Models.ipynb               # Jupyter Notebook for training models
│   │── heart.csv                  # Dataset used for training
│
│── 📂 Website
│   │── 📂 static                   # Static files (CSS, JavaScript)
│   │── 📂 templates                # HTML templates
│   │── README.txt                  # Additional project documentation
│   │── app.py                      # Flask web application
│   │── final_model.ipynb           # Final model training notebook
│   │── heart.csv                   # Duplicate dataset for website integration
│   │── model.py                    # Model inference and predictions
│   │── model15.pkl                  # Trained model - Version 1
│   │── model25.pkl                  # Trained model - Version 2
│   │── model35.pkl                  # Trained model - Version 3
│
│── .gitignore                       # Ignore unnecessary files
│── README.md                        # Project documentation

---

🔍 How It Works

1. User Inputs Health Data: The user provides details like age, cholesterol levels, heart rate, etc.
2. Machine Learning Model Predicts Outcome: The selected neural network or classification model processes the input.
3. Results Displayed Instantly: The prediction appears on the webpage, indicating whether the user is at risk for heart disease.
4. Automated Testing Mode: Users can run multiple test cases automatically to validate model accuracy.

---

📊 Model Performance & Evaluation

• Logistic Regression Accuracy: ~84.5%
• SVM Accuracy: ~84.2%
• Artificial Neural Network (ANN) Accuracy: 85.3%
• Confusion Matrices & Classification Reports provide a breakdown of precision, recall, and F1-scores.

---

🚀 Future Improvements

🔹 Enhancing Model Accuracy: Exploring deeper neural networks with more layers.
🔹 Adding More User Inputs: Incorporating additional risk factors for better prediction.
🔹 Deploying on Cloud: Hosting on AWS/GCP for public access.
🔹 Implementing a Mobile App Version 📱.

---

📜 License

This project is licensed under the MIT License.

---