Building an app for ECG data analysis

Abstract

As COVID-19 affects different people in different ways. Most infected people will develop mild to moderate illness and recover without hospitalization. Most common symptoms: • fever. • cough. • tiredness. • loss of taste or smell. A cough is usually reflex action to clear dust, phlegm, and other irritants from your lungs and windpipe. So using this cough audio and patient details. I want to predict healthy, symptomatic people without COVID-19 diagnosis and COVID-19 status. After preprocessing the dataset, I explore different models like Logistic regression, Kmean, ANN, and SVM with the dataset. The best result comes up with the ANN model, and using this, I’m able to achieve 94% accuracy in Testing Set. But For the deployable model, I have Eight different models with two different ANN architectures. One ANN has more dense than the other to extract more features from input. The prediction comes up by averaging all class predictions.