Cardiovascular and Diabetes Diseases Classification Using Ensemble Stacking Classifiers with SVM as a Meta...

Area of Science:

• Medical Informatics
• Machine Learning in Healthcare
• Computational Biology

Background:

• Cardiovascular disease (CVD) and diabetes are major global health issues with significant mortality and morbidity.
• CVD includes coronary artery diseases (CAD) and coronary heart diseases (CHD), linked to plaque buildup.
• Diabetes can lead to severe complications including heart disease, stroke, blindness, and kidney failure.

Purpose of the Study:

• To develop an improved machine learning approach for the accurate diagnosis of diabetes and cardiovascular disease.
• To address the limitations of existing classifiers with poor accuracy in diagnosing these conditions.

Main Methods:

• Development of an ensemble machine learning model named 'Stacking Classifier'.
• Utilized individual classifiers including Naive Bayes, K-Nearest Neighbor (KNN), Linear Discriminant Analysis (LDA), and Decision Tree (DT).
• Employed Random Forest and Support Vector Machines (SVM) as meta-classifiers within the ensemble.

Main Results:

• The Stacking Classifier achieved a high accuracy of 0.9735% for diabetes diagnosis, outperforming individual models (e.g., Naive Bayes at 0.7646%).
• For cardiovascular disease diagnosis, the Stacking Classifier reached 0.8871% accuracy, surpassing existing methods (e.g., SVM at 0.8472%).

Conclusions:

• The proposed Stacking Classifier demonstrates superior performance in diagnosing both diabetes and cardiovascular disease.
• Ensemble methods offer a promising direction for enhancing diagnostic accuracy in complex medical conditions.
• Further research is warranted to refine and implement these advanced diagnostic tools.