A Robust Framework for Epidemic Analysis, Prediction and Detection of COVID-19

Abstract

Covid-19 has become a pandemic that affects lots of individuals daily, worldwide, and, particularly, the widespread disruption in numerous countries, namely, the US, Italy, India, Saudi Arabia. The timely detection of this infectious disease is mandatory to prevent the quick spread globally and locally. Moreover, the timely detection of COVID-19 in the coming time is significant to well cope with the disease control by Governments. The common symptoms of COVID are fever as well as dry cough, which is similar to the normal flu. The disease is devastating and spreads quickly, which affects individuals of all ages, particularly, aged people and those with feeble immune systems. There is a standard method employed to detect the COVID, namely, the real-time polymerase chain reaction (RT-PCR) test. But this method has shortcomings, i.e., it takes a long time and generates maximum false-positive cases. Consequently, we necessitate to propose a robust framework for the detection as well as for the estimation of COVID cases globally. To achieve the above goals, we proposed a novel technique to analyze, predict, and detect the COVID-19 infection. We made dependable estimates on significant pandemic parameters and made predictions of infection as well as potential washout time frames for numerous countries globally. We used a publicly available dataset composed by Johns Hopkins Center for estimation, analysis, and predictions of COVID cases during the time period of 21 April 2020 to 27 June 2020. We employed a simple circulation for fast as well as simple estimates of the COVID model and estimated the parameters of the Gaussian curve, utilizing a parameter, namely, the least-square parameter curve fitting for numerous countries in distinct areas. Forecasts of COVID depend upon the potential results of Gaussian time evolution with a central limit theorem of data the Covid prediction to be justified. For gaussian distribution, the parameters, namely, extreme time and thickness are regulated using a statistical Y2 fit for the aim of doubling times after 21 April 2020. Moreover, for the detection of COVID-19, we also proposed a novel technique, employing the two features, namely, Histogram of Oriented Gradients and Scale Invariant Feature Transform. We also designed a CNN-based architecture named COVIDDetectorNet for classification purposes. We fed the extracted features into the proposed COVIDDetectorNet to detect COVID-19, viral pneumonia, and other lung infections. Our method obtained an accuracy of 96.51, 92.62, and 86.53% for two, three, and four classes, respectively. Experimental outcomes illustrate that our method is reliable to be employed for the forecast and detection of COVID-19 disease.

Deep learning techniques for detection and prediction of pandemic diseases: a systematic literature review

“… Face mask dataset from Masked Faces (MAFA) ResNet50V2 NA Precision 0.97, Recall 0.97, F1-score 0.97, Inference Time (ms) 7, Training Accuracy 91.93 and Validation Accuracy 90.49% Four DL-based techniques; ResNet50V2, VGG19, MobileNetV2 and InceptionV3 were used in the study for face mask detection. ResNet50V2 results appear good but better results can be achieved [ 51 ] The DL-based method for COVID-19 pandemic detection was proposed and Transfer Learning (TL) concept was incorporated Chest X-ray images dataset was used ResNet-based model NA ResNet model achieved an accuracy of 97% The outcome of ResNet appear good but better results may be achieved when optimizer is used [ 47 ] A novel method to analyze, predict, and detect the COVID-19 pandemic was proposed. Publicly available COVID-19 Chest X-ray dataset from Johns Hopkins University was used CNN-model NA The outcome of the study was accuracy of, 96.51%, precision 97.67%, recall 97.73% and F1-score 97.20 For quick and easy forecasts of the COVID-19 infection, the study used the normal distribution.…”

An AI healthcare ecosystem framework for Covid-19 detection and forecasting using CronaSona

“…Hassan et al [ 10 ] proposed a robust technique aimed at analyzing, predicting, and detecting COVID-19. This technique encompasses two primary tasks: infection forecasting and COVID detection.…”

Prediction of cross-border spread of the COVID-19 pandemic: A predictive model for imported cases outside China

“…To predict the epidemic situation of COVID-19 in a timely, accurate, and reliable manner, scholars have conducted numerous studies on the prediction, prevention and control of COVID-19 transmission [9][10][11][12][13], and an infectious disease dynamics model has been proposed. As a tool aimed at epidemic prediction and as well as actual application, this model considers the transmission speed, transmission mode, and various prevention and control measures of infectious diseases as well as other factors as a whole [14], and thus has significant application value for early warning of infectious diseases as well as for assessing prevention and control effects on the diseases.…”