🫀 Heart Disease Prediction System

Welcome to the Heart Disease Prediction System! This repository contains a complete end-to-end Machine Learning pipeline that predicts the likelihood of a patient having heart disease based on their clinical test results.

---

🚀 1. What Actually I'm Doing

I am building an automated, command-line-based predictive diagnostic system. By taking in various physiological and clinical parameters of a patient (such as age, blood pressure, cholesterol, and heart rate), this system processes the data and predicts whether the patient is at a High Risk or Low Risk of having heart disease.

🛠️ 2. How I'm Doing It

This project is implemented using Python and standard data science libraries (pandas, scikit-learn, joblib).

• Data Preprocessing: The data is cleaned (dropping duplicates) and numerical features are standardized using StandardScaler.
• Modeling: I use a Machine Learning classification algorithm. Hyperparameter tuning is applied to find the absolute best model parameters.
• Workflow: The project is split into two straightforward scripts:
  • train.py: For training the model and saving the serialized artifacts.
  • predict.py: An interactive script for feeding new patient data and receiving an instant diagnostic report.

💡 3. Why I'm Doing It

Heart disease is one of the leading causes of mortality globally. Early detection is crucial for effective treatment and survival. I am building this project to:

• Provide an accessible and fast preliminary diagnostic tool.
• Assist healthcare professionals by offering data-driven insights.
• Demonstrate how Machine Learning can be practically applied to real-world healthcare problems.

---

📊 4. About the Dataset

The dataset contains clinical and non-invasive test results of various patients.

• Download Source: UCI Machine Learning Repository - Heart Disease Dataset
• Number of Rows (Instances): 1025
• Number of Columns: 14 (13 features + 1 target variable)

Features Include:

1. age: Age in years
2. sex: Sex (1 = male; 0 = female)
3. cp: Chest pain type (0-3)
4. trestbps: Resting blood pressure (in mm Hg)
5. chol: Serum cholesterol in mg/dl
6. fbs: Fasting blood sugar > 120 mg/dl (1 = true; 0 = false)
7. restecg: Resting electrocardiographic results (0-2)
8. thalach: Maximum heart rate achieved
9. exang: Exercise-induced angina (1 = yes; 0 = no)
10. oldpeak: ST depression induced by exercise relative to rest
11. slope: The slope of the peak exercise ST segment (0-2)
12. ca: Number of major vessels (0-4) colored by fluoroscopy
13. thal: Thalassemia (0-3)
14. target: Presence of heart disease (1 = Disease, 0 = No Disease)

---

📈 5. Exploratory Data Analysis (EDA)

Understanding the data is key. In our Jupyter Notebook (Heart_Disease.ipynb), various visualizations were created to understand feature relationships:

🔹 Correlation Heatmap

The heatmap shows how strongly each feature correlates with the target. Features like cp (chest pain), thalach (max heart rate), and slope show positive correlations, while exang and oldpeak show strong negative correlations.

🔹 Histograms & Data Distribution

Histograms help us understand the age distribution, blood pressure spread, and cholesterol levels of our patient data.

🔹 Boxplots (Outlier Detection)

Boxplots were utilized to visualize continuous variables (trestbps, chol, thalach) and identify potential outliers before scaling the data.

🔹 Feature Importances

This visualizes which clinical tests are the most critical in determining the presence of heart disease.

🔹 Diagnostic Prediction Report

The terminal application generates a highly readable and intuitive Diagnostic Report detailing the calculated probabilities for the patient.

---

🧠 6. Which Model I Used and Why

Model Chosen: Support Vector Machine (SVM)

Why SVM?

1. High Dimensionality & Tabular Data: SVMs perform exceptionally well on structured, tabular data where boundaries between classes (disease vs. no disease) are relatively distinct.
2. Robustness to Overfitting: By optimizing the C (regularization) parameter, SVM avoids overfitting, especially on moderately sized datasets like ours.
3. Flexibility via Kernels: Using GridSearchCV, the pipeline automatically tests both linear and rbf (Radial Basis Function) kernels to find the most accurate non-linear hyperplanes separating the data.
4. Probabilistic Outputs: I configured the SVM to output probability percentages (probability=True), allowing us to provide detailed confidence metrics in the prediction report rather than just a simple Yes/No.

---

💻 7. How to Train and Predict (Quickstart)

Ensure you have a virtual environment set up and the required dependencies installed (pandas, scikit-learn, joblib).

Step 1: Train the Model

Run the training script to load the data, scale features, tune the SVM, and save the final models to the models/ directory.

python train.py

What happens?

• Drops duplicate records to prevent data leakage.
• Scales continuous variables (age, trestbps, chol, etc.).
• Performs cross-validation grid search to find the absolute best SVM.
• Saves scaler.joblib and svm_model.joblib.

Step 2: Make Predictions

Run the interactive prediction script to enter a patient's clinical data and get a diagnosis.

python predict.py

What happens?

• You will be prompted to enter values for all 13 clinical features.
• The script automatically scales the input based on the saved scaler.
• Evaluates the patient against the SVM model.
• Prints a structured Diagnostic Prediction Report with exact probabilities.