A Dense Network Approach with Gaussian Optimizer for Cardiovascular Disease Prediction

The effective method for cardiovascular disease (CVD) risk prediction is done by training the deep neural networks on the well-defined training dataset. The irregular subset from the real dataset with a greater data variance is considered for prediction. The proposed system uses the trained datasets to separate common and greatly biased subsets for accurately implementing the prediction models when many previous models are learning from the random samples of training datasets. The feature selection is done with a Binary Krill Herd meta-heuristic optimizer (B-KHA), and the extracted features are fed to the CapNet model for prediction purposes. In addition, the isolated training groups learn the network classifiers. This proposed model used the Cleveland dataset gathered from online resources. The experiment proves that the proposed model improves the network performance by appropriate prediction. The suggested model shows that the experimental outcomes perform better than the traditional machine learning algorithms, with 95% accuracy, 94% specificity, 94% precision, 97% sensitivity, 95% F1-score, and 90% Mathews' Correlation Coefficient (MCC).

Automated summary

This study proposes a method for cardiovascular disease risk prediction using deep neural networks, with feature selection performed by B-KHA. The experimental results show improved prediction performance compared to traditional machine learning algorithms.