Coronary Heart Disease Prediction

Overview

This project aims to predict the 10-year risk of future coronary heart disease (CHD) in patients. Heart disease has become a leading cause of death, and accurate prediction models can assist diagnosticians in reducing misdiagnosis. The project involves a thorough analysis of factors such as blood pressure, glucose levels, heart rate, and other relevant features from a dataset containing almost 4000 rows.

Problem Statement

The primary problem is to predict whether a patient has a 10-year risk of future coronary heart disease. This is crucial in the context of the rising prevalence of heart diseases and the need for accurate diagnostic tools to identify at-risk individuals.

Importance

• Prevalence of Heart Disease: Heart disease is a leading cause of mortality worldwide, underscoring the importance of accurate prediction models.

• Reducing Misdiagnosis: Reliable predictions can assist healthcare professionals in identifying patients at risk, leading to more effective preventive measures and treatments.

Dataset

The dataset comprises information on various factors, including smoking habits, blood pressure, glucose levels, gender, and more, collected from nearly 4000 individuals.

Key Features

• Smoking Habits
• Blood Pressure (Systolic and Diastolic)
• Glucose Levels
• Gender

Approach

The solution involves a comprehensive analysis of the dataset, utilizing machine learning algorithms to predict the risk of coronary heart disease based on the identified features.

Insights

• Smoking Habits: Smokers, particularly those smoking 15-20 cigarettes a day, are more prone to heart disease by 7.27% compared to non-smokers.
• Gender: Men are 6.09% more prone to heart disease than women.
• Blood Pressure: Systolic blood pressure is more useful in identifying CHD, with different levels indicating varying risks.
• Glucose Levels: Individuals with CHD often exhibit glucose levels in the range of 72-78 mg/dl.

Machine Learning Models

The project utilizes several machine learning algorithms, each evaluated for its performance in predicting CHD:

• Decision Tree Classifier: Strong performer with high accuracy and precision.
• K-Nearest Neighbors (KNN): Demonstrates good recall but needs improvement in precision.
• Random Forest Classifier: Robust choice, delivering high accuracy and precision, but watch for potential overfitting.
• Logistic Regression and SVM: Moderate performance, with room for improvement in accuracy and precision.
• Gaussian NB Classifier: Shows lower recall and overall moderate performance.

Run the Jupyter notebook for detailed analysis and model training

• coronary_heart_disease_prediction.ipynb