Heart Risk Disease Classification

> Load and explore the 'heart.csv' dataset to understand its structure and contents.

The 'heart.csv' dataset contains 1025 rows and 14 columns, comprising mostly integer data types with one float column. It includes attributes related to patient details and heart conditions such as age, sex, cholesterol levels, and target (presumably an indicator for heart disease). All columns have non-null values.

> Analyze the correlation between features in the heart dataset. Implement a basic classification model using the heart dataset.

The RandomForestClassifier achieved excellent performance on the dataset, with 98% overall accuracy and high precision, recall, and F1-scores for both classes.

The scatterplot maps the clustered data points based on the scaled features, with the K-Means algorithm identifying three clusters that likely represent varying heart disease risk levels. Cluster centroids (red markers) signify the average position for each risk group. The distribution of data points within and around these centroids offers insights into group distinctions in heart-related attributes.

This visualization plots clusters with respect to 'age' and 'cholesterol levels'. Each cluster shows how age and cholesterol interact to define heart disease risk groups. For example, clusters may reveal that individuals with high cholesterol in certain age ranges are grouped in a higher-risk category, emphasizing cholesterol as a critical risk factor. Further analysis can enhance understanding of age-cholesterol relationships in HD risk.

> Generate a detailed analysis on the significance of clustering results for heart disease risk.

The clustering results for heart disease risk provide insightful segmentation of patients into three primary groups (clusters 0, 1, and 2). Here's a detailed significance analysis:

1. Cluster Sizes: - Cluster 0: 396 individuals. - Cluster 1: 287 individuals. - Cluster 2: 342 individuals. The distribution of cluster sizes is relatively balanced, offering a proportional representation of different risk patterns.

2. Mean Feature Values by Cluster: - Cluster 0: Represents younger individuals (mean age of ~47.8), mostly males (~86.6%), with relatively normal or low cholesterol levels (mean ~227.9). These individuals exhibit higher maximum heart rates (mean ~164.98) and lower occurrences of exercise-induced angina (~12.6%). This cluster likely indicates a lower-risk group or individuals with generally healthier heart profiles. - Cluster 1: Comprising individuals with a higher average age (~59.4) but a balanced distribution of males (~32%). Cholesterol levels are elevated (~259.8), suggesting a moderate-risk group. Notably, occurrences of exercise-induced angina are higher (mean ~15.3%), reflecting strain under physical activity. - Cluster 2: Represents older individuals (mean age ~57.9), predominantly males (~80.9%). Cholesterol levels are high (~255.3), with significantly lower maximum heart rates (~129.8), higher scores for exercise-induced angina (~73.4%), and elevated oldpeak values (~1.99). This cluster likely reflects individuals with higher heart disease risk.

3. Health Risk Implications: - Cluster 0 is indicative of a relatively low-risk demographic. - Cluster 1 represents moderate-risk individuals where early medical intervention (lifestyle changes or monitoring) might significantly reduce further complications. - Cluster 2 highlights a high-risk group needing immediate attention, including advanced medical tests and care.

These findings create a foundation for developing targeted healthcare strategies and preventive measures by improving early diagnosis and understanding individual risk levels.

This visualization uses 'thal' (thallium test results) and 'oldpeak' (ST depression induced by exercise relative to rest) to display the clusters. It highlights how specific medical test results contribute to risk stratification for heart disease. The clusters indicate varying risk patterns, potentially emphasizing higher 'oldpeak' values and specific 'thal' categories as markers for increased heart disease risk.

This boxplot depicts cholesterol levels across the clusters, indicating differences in average levels and variability. Observing higher cholesterol values in certain clusters aligns with their classification as higher-risk groups for heart disease. Additional patterns in cholesterol distributions across clusters provide deeper insights into risks.

This histogram visualizes age distribution within each cluster, showing how different age groups align with heart disease risk categories. Patterns, such as a cluster predominantly consisting of older individuals, can indicate age as a significant factor in defining risk levels.

This pairplot displays relationships between key features ('age', 'chol', 'thalach', and 'oldpeak') across clusters. It highlights how risk groups differ in feature combinations, such as low 'thalach' paired with high 'oldpeak' in certain clusters signaling elevated risks. These correlations can guide targeted interventions.

> visualizations comparing cluster features and potential risk factors for heart disease.

This barplot visualizes the distribution of gender (male and female) within the clusters. It appears that males dominate certain clusters, potentially signaling higher risk factors related to specific heart disease markers.

> visualizations to compare age and cholesterol levels across heart disease risk clusters.

The scatter plot with regression lines compares age and cholesterol levels across clusters, illustrating trends within each risk group. It highlights how cholesterol changes with age and risk category differences in age-related cholesterol patterns.

The IEEE-format research paper "Heart_Disease_Clustering_Research_Paper.docx" has been successfully created. Let me know if you need additional edits or enhancements.

Created in Deepnote