Prediction of COVID-19 for diabetes patients using pre-trained convoluted Recurrent Neural Network and fast 3D-convolution neural network with U-net++

Abstract

This paper aims to predict COVID- 19 to analyse the death rate due to the presence of diabetes. At present, the entire world witnessed the rapid and severe spread of theNovel Coronavirus (COVID-19) since 2019. This spread throughout mankind and animals. Its severe complication and without the earlier disease detection and treatment leads to death. So CAD-based system is obtained for rapid detection of disease and predict the death rate due to diabetes which involves machine learning techniques. First to collect the real-time public dataset of COVID-19 patients with diabetes topredict thedeath rate due to thepresence of diabetes. Here the data has been collected through HDFS. Initially, the diabetes data and chest X-ray for COVID patients has been collected and classified for obtaining the abnormal range of diabetes in corona patients. Here for classification, we use apre-trained convoluted recurrent neural network (PCRNN) and fast 3D-convolution neural network with U-net++ (F-3D-CNN-U net++) which use for classifying the numerical data and image data. The numerical data has been classified using pre-trained convoluted recurrent neural network and image data has been classified using afast 3D-convolution neural network with U-net++. By diagnosing the diabetes range and coronavirus affected part of thelungs, the death rate has been calculated. The experimental results show accuracy, precision, recall, F-1 score, true positive, false positive rate, MSE, MLR.