Heart Disease Prediction

Catching heart disease before it’s too late.

Cardiovascular disease remains one of the leading causes of mortality worldwide. Early prediction using patient health data can significantly improve prevention and treatment outcomes. This project builds and evaluates three ML classification models—Logistic Regression, Decision Tree, and Random Forest—for CVD prediction, emphasizing recall to minimize missed diagnoses.

Data-driven prevention, not reaction.

The project compares traditional and ensemble approaches on a real CVD dataset, with GridSearchCV hyperparameter tuning and a focus on reducing false negatives—the most critical metric in healthcare ML.

End-to-end ML pipeline.

The full pipeline: data loading, EDA, preprocessing (encoding, scaling, handling missing values), model training with hyperparameter tuning, and evaluation using Accuracy, Precision, Recall, F1-Score, and Confusion Matrix.

Minimizing false negatives in healthcare.

In medical diagnosis, a false negative (missing a sick patient) is far more dangerous than a false positive. The challenge was tuning three models — Logistic Regression, Decision Tree, Random Forest — to maximize recall without destroying precision. Random Forest with tuned thresholds struck the best balance.