Artificial Neural Network for Cardiovascular Disease Prediction Model based on Feature Subset Correlation Analysis
Backgrounds: Cardiovascular Disease (CVD) is one of the leading causes of death globally and it is preventable but it progresses in silence. Also in Korea, CVD was the most dangerous disease which led people to death for decade except cancer. Researches about diagnosing CVD using Artificial Intelligence (AI) and Machine Learning (ML) are progressing recently.
Problems: It is hard to predict CVD because of the difficulty of analyzing correlated features about causes of CVD. Moreover, differences of the characteristic of the races requires another prediction method that fits to them.
Methods: In this paper, we suggest Artificial Neural Network (ANN) learning for the prediction model of CVD based on feature subset correlation analysis. Our method is selecting the features that is strongly related to CVD from the Korea National Health and Nutrition Examination Survey (KNHNES) – VII dataset by evaluating and comparing the ANNs which are learned from the feature subsets. Finally, we provide a CVD prediction ANN which are learned from the selected features.
Conclusion: The CVD prediction model that is implemented by the method could predict CVD of specific race group and find important features that are related to CVD. Also, we could prevent CVD by prediction.
Index Terms - Cardiovascular Disease, Feature Selection, Feature Subset Correlation, Machine Learning, Artificial Neural Network, KNHNES-VII