Heart Disease Prediction Project

Overview

This project aims to predict the presence of heart disease using the Hungarian subset of the UCI Heart Disease dataset. We focus on three key predictors: age, cholesterol levels, and maximum heart rate. The goal is to determine the effectiveness of these predictors in forecasting heart disease using a k-nearest neighbors (KNN) classification model.

Dataset

• Source: UCI Heart Disease dataset (Hungarian subset)
• Features Used: Age, Cholesterol, Maximum Heart Rate
• Target: Presence of Heart Disease

Methodology

1. Data Preprocessing:

   • Cleaned the dataset by replacing missing values and converting data types.
   • Split the data into training (75%) and testing (25%) sets.

2. Model Development:

   • Standardized the data.
   • Applied k-nearest neighbors (KNN) classification.
   • Used cross-validation to determine the optimal number of neighbors (k).

3. Model Evaluation:

   • Evaluated the model on the test set using accuracy, precision, recall, and confusion matrix.

Results

• Optimal k: 13
• Accuracy: 65.2%
• Precision: 68.6%
• Recall: 81.4%

Dependencies

• R
• Libraries: tidyverse, repr, tidymodels, gridExtra, cowplot

How to Run

Open the provided R file "project_report.ipynb".

References

• Di Cesare M, et al. (2024). The Heart of the World. Glob Heart.
• Harvard Health. (2023). What is a normal heart rate?
• Janosi, A., et al. (1988). Heart Disease. UCI Machine Learning Repository.
• Rodgers JL, et al. (2019). Cardiovascular Risks Associated with Gender and Aging. J Cardiovasc Dev Dis.