Empirical Analysis of Hybrid Machine Learning based Earlier Stage Covid Disease Prediction using Lung CT Images

Corona Virus or Covid-19 Disease, a term which creates a mass affection to all the countries in both Human life as well as economy wise. This disease causes a huge destruction in many person's individual life and most of the people around the world died due to this cause. Several assessments and researches are going on to predict the disease in fine manner as well as identify the disease over earlier stages to save the life of people without any delay. However, the most prominent and acceptable way of predicting the Corona Virus is by using the Lung Computed Tomography (CT) images. The Lung based CT images provides a huge support to identify the Covid virus on earlier stages, in which the people are advised to take such type of scanning while infected with corona virus. An earlier stage identification of Corona Virus is the basic need now-a-days, in which the disease is identified initially, means it can easily be cured. The identification of Covidvirus over lung CT images is of course a complex task because the CT images contains low-intensity pixels and the contrast level variations are different on various images. So it is complex to manipulate such images in practical, due to this a novel Digital Image Processing scheme is required to provide an efficient support to the respective physician to identify the Corona Virus on earlier stages in clear manner. The concept of machine learning is adopted over this paper to provide a proper predictions as well as the logic of dual classification algorithms are combined together to form a new machine learning strategy to attain high accuracy with enhanced prediction probabilities. The logic of Deep Neural Network (DNN) is modulated with respect to the logic of Random Forest (RF) Classification algorithm to make a new methodology called Hybrid Learning based Disease Prediction Scheme (HLDPS). In which this proposed approach associates the benefits of both DNN and RF into this prediction strategy to make an appropriate predictions over lung CT images and report the level of severity based on the cell vector distance. The resulting section of this paper provides proper experimental proof of the mentioned things in clear manner with graphical representations. For all the proposed approach of HLDPS is sufficient to predict the Corona Virus on earlier stages based on lung CT images in fine manner and the associated proofs are specified clearly on resulting section of this paper.