Machine Learning Methods for Hypercholesterolemia Long-Term Risk Prediction.

Area of Science:

• Cardiology
• Biomedical Informatics
• Data Science

Background:

• High cholesterol (hypercholesterolemia) poses significant risks to heart health by blocking blood vessels and impeding circulation.
• Machine learning (ML) offers powerful tools for risk prediction in healthcare, assisting medical professionals.
• Early and accurate prediction of hypercholesterolemia is crucial for preventing severe cardiovascular events.

Purpose of the Study:

• To develop efficient machine learning (ML) risk prediction tools for hypercholesterolemia.
• To identify key features associated with hypercholesterolemia through data analysis.
• To evaluate and compare the performance of various ML models for hypercholesterolemia risk prediction.

Main Methods:

• A supervised machine learning (ML) methodology was employed.
• Data understanding analysis was performed to explore feature importance for hypercholesterolemia.
• Multiple ML models were trained and tested, with performance evaluated using precision, recall, accuracy, F-measure, and AUC metrics.

Main Results:

• Soft Voting with Rotation and Random Forest models demonstrated superior performance.
• These models achieved an AUC of 94.5%, precision of 92%, recall of 91.8%, and F-measure of 91.7%.
• An overall accuracy of 91.75% was recorded for the best-performing models.

Conclusions:

• The developed ML models, particularly Soft Voting with Rotation and Random Forest, are highly effective for hypercholesterolemia risk prediction.
• These findings can support healthcare providers in proactive patient management and cardiovascular disease prevention.
• The study underscores the potential of ML in enhancing diagnostic accuracy and risk stratification for hypercholesterolemia.