CardioGuardHeart Health Predictor

Compare Heart Risk Prediction Models

Explore the performance of different machine learning models used for heart attack risk prediction.

Model Comparison Mode

Compare models trained for different prediction tasks

Multiclass Models: These models predict heart disease severity on a scale from 0 (No Disease) to 4 (Critical), providing more granular risk assessment but typically with lower overall accuracy.

Multiclass Model Performance

Visual comparison of multiclass model performance metrics

Multiclass Model Details

5-Class Prediction
Random Forest
An ensemble learning method that operates by constructing multiple decision trees during training.
Multiclass

Accuracy

74.0%

Precision

72.0%

Recall

74.0%

F1-Score

73.0%

Key Features

Handles non-linear relationshipsRobust to outliersHigh performance on complex datasets

Best For

Complex classification tasks with many features

Limitations

Shows weakness on minority classes as seen in the classification report

Gradient Boosting
An ensemble technique that builds models sequentially, with each new model correcting errors from previous ones.
Multiclass

Accuracy

70.0%

Precision

70.0%

Recall

70.0%

F1-Score

70.0%

Key Features

High performanceHandles mixed data typesSequential learning approach

Best For

Structured data problems with numerical and categorical features

Limitations

Requires careful tuning to prevent overfitting

XGBoost
An optimized distributed gradient boosting library designed for efficient and scalable training.
Multiclass

Accuracy

68.0%

Precision

69.0%

Recall

68.0%

F1-Score

69.0%

Key Features

Regularization to prevent overfittingHandles missing valuesParallel processing

Best For

Large datasets with complex relationships

Limitations

Can be computationally intensive and requires parameter tuning

SVC
Support Vector Classification finds the hyperplane that best separates classes in a high-dimensional space.
Multiclass

Accuracy

67.0%

Precision

63.0%

Recall

67.0%

F1-Score

64.0%

Key Features

Effective in high dimensional spacesMemory efficientVersatile through different kernel functions

Best For

Classification with clear margins between classes

Limitations

Performance drops when classes overlap significantly