Tri-Hybrid Naive Bayes Classification Model for Early Cardiovascular Disease Detection

Cardiovascular disease is a major global health problem affecting people around the globe. Although modern medicine has contributed in providing significant data to help mitigate cardiovascular disease, however, there is still needs to provide an ideal solution that will help in detecting early cardiovascular diseases. As a result, existing researchers have proposed several hybrid cardiovascular disease detection techniques using both machine learning and deep learning approaches. However, most solution requires improvement especially in term of accuracy detection. In this paper, a tri-hybrid Naïve Bayes classification algorithm for accuracy prediction of early cardiovascular disease detection is developed. In the developed algorithm, the strength of the conventional Naïve Bayes classification algorithm is improved with that of decision tree approach to achieve better accuracy. To further enhance its accuracy, Neural Network (NN) procedure is later incorporated into the improved Naïve Bayes classification algorithm. Implementation of the developed tri-hybrid algorithm is carried out on Waikato Environment for Knowledge Analysis (WEKA) and evaluated using dataset from the kaggle website that contain instances of 1025 and 14 attributes. Experimental results based on confusion matrix for performance evaluation indicates the developed tri-hybrid Naïve Bayes algorithm has achieved high classification accuracy and recall rates of 98.54 and 99% respectively when compared with that of the benchmarked schemes. The performance of the proposed solution deployed can help cardiologist make better prediction in the diagnosis of the heart disease. Although several machine learning algorithms as seen in the literatures are trained on historical data, these algorithms may not be able to accurately predict the risk of heart disease in patients who are exposed to new or emerging risk factors.  Further research is to experiments a wider range of algorithms with a build in model that can detect early cardiovascular diseases as well improve accuracy.