A Framework for Diagnosis of Covid-19 Infection using Deep Learning Approach

Abstract: Coronavirus is a respiratory sickness that is impelled by a novel Covid. The basic indications show up in the contaminated individual are fever, cough, sore throat, and trouble in relaxing. Disappearing of taste, sluggishness, throbs, and nasal blockage can likewise be seen in certain patients. The length among tainting and the principal sign of manifestations might be reached out to 14 days [4]. The disease of this infection is communicated through the beads of patients, for example, coughing and sniffling. In the event that the individual comes by implication or in a roundabout way contact with a contaminated individual, at that point the reached individual gets tainted. The antibodies/medications of this sickness are not accessible as of not long ago. Segregation and social distancing are the lone answers for this disease. Hence, the early recognition of tainted people is needed to stop the spread of contamination. This paper presents a framework for diagnosis of covid-19 infection using Deep Learning approach. The proposed system starts by making use of a covid-19 dataset, which is made up of 6 columns and 48 rows. The dataset comprises of most covid-19 symptoms ranging from dry cough, sore throat, high fever and difficulty in breathing of 48 patients and also the results which shows if the patients is infected with covid-19 or not. We made use of a feed forward neural network in training our model and we had an accuracy of about 92%. The trained model was saved and deployed to web using python flask so that users can enter in most covid-19 symptoms and check if they have been tested positive to the virus or not.

Keywords: Deep Learning, Covid-19, Feed Forward Neural Network, Diagnosis System