A Novel Deep Learning Model for Predicting Severity of the Disease using Covid-19 Long Non-Coding RNA Sequence.

Abstract

The Covid-19 is very critical situation throughout the world from December 2019. In this pandemic situation some of the patients are symptomless and others have many varying symptoms of the disease. The severe symptom cases are leading to damage of multi organ and death. The mechanism used to indicate severity of the disease is still unknown. Early detection of the infected patients of Covid-19 is very important to control the spreading of diseases and also severity level of the disease helps to decrease the mortality rate. So deep learning model is developed to predict the severity of the Covid-19 disease by using lncRNA sequence which is a new approach. To train this deep learning model 3363 lncRNA mutation sequences of Covid-19 patients is used. Severity is predicted by the risk score of this model which indicates whether the patient suffering with severe and non-severe disease level. The performance is evaluated by comparing this model with other three Deep Learning algorithms i.e., LSTMs, RNN and GANs. The performance of this proposed model with CNN algorithm is best among all other algorithms and also it is able to predict the severity of the disease with high accuracy prediction of 96%, which helps to reduce mortality rate