Bidirectional Recurrent Network and Neuro-fuzzy Frequent Pattern Mining for Heart Disease Prediction

Abstract

In recent days medical clinics need more analysis in heart disease because of most dangerous disease caused mostly worldwide affected by the people. By analyzing the characteristic features are high dimension due to complex structure of data analysis. Hence, the input features can be extracted with deep learning (DL) techniques to impart relevant recommendations and forecasts. DL techniques also play an essential character in early diagnosis and keeping an eye on heart disease. Numerous types of research have been conducted in this medical domain to predict heart disease at an early stage. To this end, we propose a novel Bi-directional Recurrent Network and Neuro-Fuzzy-based (BRN-NF) frequent pattern mining for heart disease prediction. Meanwhile, in the BRN-NF method, a Bi-directional Recurrent Neural Network is used for frequent pattern (i.e., feature) mining. Next, with the mined results, the Bi-directional Recurrent Network Neuro-Fuzzy Inference algorithm is employed for heart disease prediction. We endorse and calculate the disease deficiency rate based on feature selection and classification to analyses the data from the Cardiovascular Disease data set. Experiments and comparisons on Cardiovascular Disease data show that, compared to existing heart disease prediction considering highly accurate predictors and considering present/past factors results in improvements in prediction time, prediction accuracy, sensitivity and specificity to a significant extent. The accurate heart disease predictions acquired from the comparative explorations indicate the notable performance of our proposed method.