This project focuses on predicting the likelihood of death in patients with heart failure using clinical records and a variety of machine learning algorithms. The dataset includes 299 real patient records with features like age, blood pressure, serum creatinine, diabetes, and other health indicators. By applying and comparing multiple classifiers, the goal is to identify patterns that could support early intervention and better patient outcomes.
🔍 Project Highlights
-
Data Exploration: The dataset was split into categorical and continuous features to streamline the analysis.
-
Preprocessing: Applied standard scaling and train-test splitting for model consistency.
-
Model Comparison: Tested multiple classification models including:
a) Logistic Regression b) Support Vector Classifier c) K-Nearest Neighbors d) Decision Tree e) Naive Bayes f) Random Forest
- Evaluation: Accuracy scores were compared across models to identify the most reliable predictor for mortality.
💡 Why This Project Matters Heart failure is a critical public health issue, and early prediction of patient risk can significantly improve outcomes. This project showcases how machine learning can enhance clinical decision-making by identifying high-risk individuals through data-driven modeling.
🚀 What's Unique
- A hands-on comparison of multiple algorithms on a healthcare dataset.
- Practical and explainable steps in preprocessing, making it replicable and easy to adapt for other clinical conditions.
- Clear separation of features to distinguish clinical indicators and patient demographics.
🔧 Potential Extensions
- Add model interpretability tools like SHAP or LIME to explain predictions.
- Integrate cross-validation and hyperparameter tuning.
- Deploy the best model as a web application for real-time clinical use.