Rethinking the Transfer Learning Architecture for Respiratory Diseases and COVID-19 Diagnosis

The emergence of COVID-19 has caused a disastrous scenario worldwide, becoming one of the most acute and deadly diseases in the last century wreaking havoc on the health and lives of countless people. The prevalence rate of COVID-19 is growing significantly every day across the world. One critical step in combating COVID-19 is the capacity to identify infected individuals and place them in special care as soon as possible. Detecting this condition via radiography and radiology images is one of the quickest ways to diagnose patients. Early study has found specific abnormalities in the chest radiographs of infected individuals with COVID-19. Inspired by prior research, we examine the application of transfer learning models to detect COVID-19 patients in X-rays. In this study, an X-ray image collection from patients with common bacterial pneumonia, viral pneumonia, proven COVÍD-19 disease, and normal occurrences was used to diagnose coronavirus disease automatically. A dataset has been used in this experiment comprising 76 image samples showing verified COVID-19 illness, 2786 images showing bacterial pneumonia, 1504 images showing viral pneumonia, and 1583 images showing normal circumstances. The information was gathered from publicly accessible X-ray images. Data augmentation technique is applied to the trained image dataset. Two transfer learning models, namely, VGG 16 and Xception, have been modified in this paper after applying additional layers with the base model. Modified Xception model provides an overall accuracy of 84.82% for Adam optimizer and 78.40% for RMSprop optimizer. Modified VGG 16 model provides an overall accuracy of 84.98% for Adam optimizer and 83.88% for RMSprop optimizer. In addition to accuracy, we show each model’s receiver operating characteristic (ROC) curve, precision, recall, F1-score, and AUC. © 2022, The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd.

A Modern Paradigm for Diagnosing Novel Coronavirus Disease (COVID-19) Using Multilayer Customized CNN via X-ray Images

The novel disease that has already been declared a global pandemic that is COVID-19, initially had an epidemic in a major Chinese city called Wuhan, China. This novel virus has now infected more than two hundred countries across the world as it propagates through human activity. In comparison, novel coronavirus signs are very close to general seasonal influenza such as common cold, fever, cough and shortness in breathing. Infected patient monitoring is viewed as a crucial phase in the battle against COVID-19. Detection tools for Positive cases of COVID-19 do not offers distinctive results, so that it has increased the need to support diagnostic tools. Therefore, to prevent further dissemination of this disease, it is extremely important as early as possible to identify positive cases. However, there will be some approaches for identifying positive patients of COVID-19 that are usually conducted on the basis of respiratory samples and amongst them, X-Ray or radiology images are an essential treatment course. Latest data from the techniques of X-Ray imaging show that these samples contain significant SARS-CoV-2 viruses. information. In order to reliably diagnose this virus, the use of deep learning techniques that is DNN which is also offers advanced imaging instruments and techniques will prove to be useful, as can the issue of the absence of trained rural physicians. In this report, we presented a multilayer customized convolution neural network (MC-CNN) system analyzing chest X-Ray images of individuals suffering from covid-19 using an open-source database available in kaggle. In order to propose DNN approach provides 97.36% of classification accuracy, 97.65% of sensitivity, and 99.28% of precision. Therefore, we conclude that this proposed approach will allow health professionals to confirm their initial evaluation of patients with COVID-19. © 2021, Springer Nature Singapore Pte Ltd.

Forecasting the Spread of COVID-19 and ICU Requirements

Since December 2019, the world is fighting against coronavirus disease (COVID-19). This disease is caused by a novel coronavirus termed as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). This work focuses on the applications of machine learning algorithms in the context of COVID-19. Firstly, regression analysis is performed to model the number of confirmed cases and death cases. Our experiments show that autoregressive integrated moving average (ARIMA) can reliably model the increase in the number of confirmed cases and can predict future cases. Secondly, a number of classifiers are used to predict whether a COVID-19 patient needs to be admitted to an intensive care unit (ICU) or semi-ICU. For this, classification algorithms are applied to a dataset having 5644 samples. Using this dataset, the most significant attributes are selected using features selection by ExtraTrees classifier, and Proteina C reativa (mg/dL) is found to be the highest-ranked feature. In our experiments, random forest, logistic regression, support vector machine, XGBoost, stacking and voting classifiers are applied to the top 10 selected attributes of the dataset. Results show that random forest and hard voting classifiers achieve the highest classification accuracy values near 98%, and the highest recall value of 98% in predicting the need for admission into ICU / semi-ICU units.

AN ADAPTIVE NEURO FUZZY MODELLING AND PREDICTION SYSTEM FOR DIAGNOSIS OF COVID-19

Artificial Intelligence has revolutionized medical sciences by providing effective ways of diagnosing various diseases. The main objective of this paper is to design a system which is able to diagnose possible presence of COVID-19 in a patient using Adaptive neuro fuzzy inference system (ANFIS). ANFIS is an approach which can be considered as a amalga-mation of artificial neural networks and fuzzy systems and hence providing advantages of both of them. Our proposed system functions with 5 variables as input and 1 variable as output. A comparative performance analysis of results obtained from ANFIS and fuzzy systems is also done which clearly depicts that ANFIS model outperforms fuzzy systems by achieving better accuracy than fuzzy systems for the diagnosis of COVID-19.

A Review on COVID-19

Background: Coronavirus is a family of viruses, and they are named coronavirus based on the crown-like spikes they have on their surface. The word “Corona” is a Latin word that means “crown.” Recently a virus of the corona family emerged in Wuhan, Hubei, China. On December 31, China informed WHO about some patients having unidentified pneumonia. It was initially named novel coronavirus because of its uniqueness. But later the coronavirus study group of the International Committee on Taxonomy of Viruses designated it as severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). SARS-CoV-2 has affected the entire world, infected more than a million people till now, and claimed more than 235,288 lives so far. Objective: This study aimed to present a case study of the recent research related to the coronavirus and proposed technology related to coronavirus. Its focus is on how infections can be caught as early as possible and what control measure should be taken to stop the virus from further spreading. Only scientific and mathematical models have been considered. Method: This study refers to the WHO website for credible information regarding the coronavirus. Many research papers and medical articles were studied before proceeding with this paper. The methodology proposed by the researchers has been mentioned in this paper. © 2021, The Author(s), under exclusive license to Springer Nature Switzerland AG.