Detection of cardiovascular disease cases using advanced tree-based machine learning algorithms.

Area of Science:

• Machine Learning
• Cardiovascular Health
• Biostatistics

Background:

• Cardiovascular disease (CVD) is a leading cause of death and disability worldwide.
• Accurate prediction of CVD is crucial for timely intervention and prevention strategies.

Purpose of the Study:

• To identify the optimal tree-based machine learning method for cardiovascular disease (CVD) detection.
• To compare the performance of various machine learning models in predicting CVD.

Main Methods:

• Analysis of data from 9,499 participants, considering 38 variables and village as a cluster variable.
• Fitting and comparing four tree-based models: standard decision tree, random forest, Generalized Linear Mixed Model tree (GLMM tree), and Generalized Mixed Effect random forest (GMERF).
• Evaluation of models using Area Under the ROC Curve (AUC) and identification of key predictive variables.

Main Results:

• Five key variables identified for CVD prediction: age, LDL cholesterol, family history of cardiac disease, physical activity, and hypertension.
• AUC values for the models were: Decision Tree (0.56), Random Forest (0.73), GLMM tree (0.78), and GMERF (0.80).
• The GMERF model exhibited the highest predictive performance.

Conclusions:

• The Generalized Mixed Effect random forest (GMERF) model is the most effective tree-based machine learning approach for CVD prediction in this dataset.
• Accounting for data clustering is important for improving prediction accuracy and developing targeted CVD prevention frameworks.