Detection Of COVID-19 Infection Research Articles

Performance of CO-RADS in the diagnosis of COVID-19 in hemodialysis patients.

Different data systems have been used to reach a common decision in the interpretation of chest computed tomography (CT) scans for the detection of COVID-19 infection. The aim of our study was to determine the usefulness of the COVID-19 Reporting and Data System (CO-RADS) for COVID-19 in patients undergoing hemodialysis. We included 90 hemodialysis patients who underwent chest CT and had samples available for real-time reverse transcription-polymerase chain reaction (RT-PCR). The files of the patients were retrospectively reviewed and the data were recorded. Image interpretation and CO-RADS staging were performed retrospectively by two radiologists experienced in COVID-19 patients, blinded to the RT-PCR results. The RT-PCR results were then compared with the CO-RADS stages obtained. The success of CO-RADS in diagnosing COVID-19 was evaluated according to its prediction of a positive RT-PCR result. At the same time, the relationship between CO-RADS stages and prognosis was also evaluated. Patients were divided into two groups according to the RT-PCR results, of which 38 (42.2%) had positive results. Validity tests of CO-RADS were performed according to the RT-PCR test. While 71.1% of patients with CO-RADS ≥4 were RT-PCR (+), 13.2% of patients with CO-RADS ≤2 were RT-PCR (+) (p < 0.001). CO-RADS ≥4 detected COVID-19 with 71.1% sensitivity and 51.9% specificity. ROC analysis confirmed the diagnostic performance of CO-RADS for predicting RT-PCR positivity with AUC = 0.73 (95% CI: 0.63-0.84). The relationship between CO-RADS stages and mortality was not significant (p = 0.21). CO-RADS is moderately sensitive and poorly specific for the diagnosis of COVID-19 in hemodialysis patients. In these patients, CO-RADS would be more useful to exclude other infections than to diagnose COVID-19.

Retraction Note: A MobileNet-based CNN model with a novel fine-tuning mechanism for COVID-19 infection detection

Retraction Note: A MobileNet-based CNN model with a novel fine-tuning mechanism for COVID-19 infection detection

Evaluation of STANDARD™ M10 SARS-CoV-2 from bronchoalveolar lavage samples

Detection of SARS-CoV-2 in bronchoalveolar lavage (BAL) is considered as a promising alternative method to detect COVID-19 infection. STANDARD™ M10 SARS-CoV-2 assay on 150 negative and 50 positives BAL samples for SARS-CoV-2 showed 96 % sensitivity, 100 % specificity compared to Allplex™ SARS-CoV-2 assay and a 31.25 genomic copies/mL limit of detection.

DRESCNN: Deep RESNET Convolutional Neural Network Based Classification of X-Ray Images for Detection of COVID-19

In this research work a solution is proposed to address the challenges for detection of Coronavirus disease (COVID-19) Infection through an innovative healthcare framework. The proposed system leverages predictive analytics on patient information and facilitates consultation, eliminating the delay in diagnosis and treatment. The solution is modular, encompassing various health services, thereby consolidating major healthcare concerns under one unified platform. The images are the inputs for the models provided on this platform which are passed to the preprocessing algorithm followed by passing to pretrained models or an algorithm to analyze and predict the disease. Finally, classification of the images of X-ray for COVID-19 Infection detection using Deep Convolutional Neural Network Models like ResNet50, InceptionNet, MobileNet are used at a detailed level. After classifying these X-ray images with different CNN variants, a majority voting is applied for selecting more accurate class label. It is observed that ResNet50 is giving highest accuracy frequently. Hence Deep ResNet Convolutional Neural Network (DResCNN) is used. This proposed solution not only tackles the critical issue of delayed decision-making and treatment but also introduces cost-effective measures by minimizing the need for extensive hospital visits.

Sex-specific differences in physiological parameters related to SARS-CoV-2 infections among a national cohort (COVI-GAPP study).

Considering sex as a biological variable in modern digital health solutions, we investigated sex-specific differences in the trajectory of four physiological parameters across a COVID-19 infection. A wearable medical device measured breathing rate, heart rate, heart rate variability, and wrist skin temperature in 1163 participants (mean age = 44.1 years, standard deviation [SD] = 5.6; 667 [57%] females). Participants reported daily symptoms and confounders in a complementary app. A machine learning algorithm retrospectively ingested daily biophysical parameters to detect COVID-19 infections. COVID-19 serology samples were collected from all participants at baseline and follow-up. We analysed potential sex-specific differences in physiology and antibody titres using multilevel modelling and t-tests. Over 1.5 million hours of physiological data were recorded. During the symptomatic period of infection, men demonstrated larger increases in skin temperature, breathing rate, and heart rate as well as larger decreases in heart rate variability than women. The COVID-19 infection detection algorithm performed similarly well for men and women. Our study belongs to the first research to provide evidence for differential physiological responses to COVID-19 between females and males, highlighting the potential of wearable technology to inform future precision medicine approaches.

A Relative Analysis of Different CNN Based Models for COVID-19 Detection using CXR and CT Images

COVID-19 is an extremely contagious disease that transmits from person to person by contaminated droplets or virus containing airborne particles. It causes severe damage to a patient’s lungs by forming patchy pulmonary lesions and consolidations, which are apparent in chest radiographs such as CXR (X-ray) and CT (computed tomography) images. Therefore, CXR and CT are considered crucial sources of information for early detection of COVID-19 infection. Manual inspection of this information requires expertise, high manpower and substantial amount of time. In order to tackle these issues, deep learning techniques can be utilized in the field of COVID-19 detection. This paper aims at analyzing the performance of different convolutional neural network models in COVID-19 detection using CXR and CT images. These models employ transfer learning and are formed by combining four well-known convolution bases with five distinct machine learning classifiers. All the models are comprehensively trained and tested on CXR and CT datasets each and are thoroughly compared with one another in terms of various evaluation metrics. Amongst these models, the best classification accuracy of 91.18% is provided by the Inception V3 with a neural network classifier on CXR images. Moreover, to assess the improvement of a COVID-19 detection method due to using different techniques, a comparative study of these transfer learning based models with other existing frameworks is also provided.

Robust Framework for Diagnosing Novel Corona Virus from CT images using Support Vector Binary Classifier

The novel coronavirus (commonly abbreviated as CoVID-19) has emerged as a threat to the entire global civilization and has emerged as one of the most infectious and, at times, deadly viruses. Prompt discovery of this infection can assist medical supervisors in taking preventive actions to control the spread. Usually, radiologists and medical specialists require an average time of ~31 minutes to test the CT images and confirm the infection. A large dataset of more than 1000 patients has been gathered and randomly chosen for this experiment. In this research, a ready-to-deploy computer-aided diagnosis (CADx) to detect COVID-19 infection is introduced. A variety of deep learning architectures have been experimented to discover the most reliable predictive model for the diagnosis. This research uses the Densely Connected Convolution Network (DenseNet-121 architecture) along with a boosting support vector binary classifier to tell the difference between someone who has the coronavirus and someone who is healthy. The combination put forward in this work achieved 93% ± 1.8% accuracy, 94.9% ± 2.6% recall, 98% ± 1.5% precision, and an F1 score of 94% ± 1.7%. The model takes less than 1 second to process one image. On the grounds of the above findings, it can be concluded that the proposed approach can be used to diagnose novel coronavirus infections.

Detection of COVID-19 infection from CT images using the medical photogrammetry technique

Medical data such as computed tomography (CT), magnetic resonance imaging (MRI), and Ultrasound images are used in medical photogrammetry. CT images have been used frequently in recent years for the diagnosis of COVID-19 disease, which has contagious and fatal symptoms. CT is an effective method for early detection of lung anomalies due to COVID-19 infection. Machine learning (ML) techniques can be used to detect and diagnose medical diseases. In particular, classification methods are applied for disease diagnosis and diagnosis. This study proposes traditional machine learning algorithms Random Forest, Logistic Regression, K-Nearest Neighbor and Naive Bayes, and an ensemble learning model to detect COVID-19 anomalies using CT images. According to the experimental findings, the proposed ensemble learning model produced an accuracy of 96.71%. This study can help identify the fastest and most accurate algorithm that predicts CT images with Covid-19 during the epidemic process. In addition, machine learning-based approaches can support healthcare professionals and radiologists in the diagnostic phase.

2283. COVID-19 among Febrile Patients Presenting without Respiratory Symptoms in Six Hospitals along the Thai-Lao and Thai-Myanmar Borders, 2021-2011

Abstract Background Thailand’s national surveillance system to detect COVID-19 infection typically relies on a combination of fever, respiratory symptoms, and risk factors. We describe the frequency, epidemiologic characteristics, and risk factors for COVID-19 in patients without respiratory symptoms and the utility of saliva and nasal swabs (NS) to detect SARS-CoV-2. Methods During June 2021-December 2022, a period of high COVID-19 incidence in Thailand, we enrolled patients aged &amp;gt;2 years presenting with a temperature ≥37.5 °C or a history of fever ≤14 days and no cough, dyspnea, or sputum production at 6 hospitals in Nakhon Phanom and Tak provinces, Thailand. Saliva, NS, and nasopharyngeal/oropharyngeal swabs (NP/OP) were collected to detect SARS-CoV-2 using real-time RT-PCR. Multivariable analysis was done to assess factors independently associated with COVID-19 infection. Cases were defined as persons with SARS-CoV-2-positive RT-PCR specimens. Results Cases were detected in 358/1,648 (22%) persons. The median age among cases was 31 (IQR 21-45) years; 210 (59%) were female. Compared to non-cases, cases were more likely to present with malaise (52% vs. 40%, p&amp;lt; 0.001), hyposmia (10% vs. 1.4%, p&amp;lt; 0.001), dysgeusia (11% vs. 4.8%, p&amp;lt; 0.001), and muscle pain (53% vs. 47%, p=0.04); cases were less likely to have chills (41% vs. 50%, p=0.002), low appetite (29% vs. 41%, p&amp;lt; 0.001), nausea/vomiting (17% vs. 29%, p&amp;lt; 0.001) and abdominal pain (4.7% vs. 16%, p&amp;lt; 0.001). Cases were more likely to have been in contact with persons with COVID-19 infection (adjusted odds ratio [95% confidence interval]: 12 [8.1-16]), febrile household members (2.6 [1.8-3.8]), and persons with respiratory symptoms (2.2 [1.4-3.7]), have a history of travel (1.7 [1.0-2.7]), current alcohol consumption (1.7 [1.2-2.4]) and obesity (1.5 [1.1-1.9]). Among cases, SARS-CoV-2 was detected in 95% of saliva, 92% of NS, and 99% of NP/OP specimens; RT-PCR was positive in all 3 specimen types from 319 (89%) cases. Compared to NP/OP, the sensitivity of testing saliva and NS were 96% and 93%. Conclusion SARS-CoV-2 was detected in more than one-fifth of febrile patients without respiratory symptoms. A broader national surveillance case definition would increase case detection. Saliva and NS demonstrated high sensitivity for SARS-CoV-2 detection. Disclosures All Authors: No reported disclosures

Nucleic acid based point-of-care diagnostic technology for infectious disease detection using machine learning empowered smartphone-interfaced quantitative colorimetry

We report a nucleic acid-based point of care testing technology for infectious disease detection at resource limited settings by integrating a low-cost portable device with machine learning-empowered quantitative colorimetric analytics that can be interfaced via a smartphone application. We substantiate our proposition by demonstrating the efficacy of this technology in detecting COVID-19 infection from human swab samples, using the RT-LAMP protocol. Comparison with gold standard results from real-time PCR evidences high sensitivity and specificity, ensuring simplicity, portability, and user-friendliness of the technology at the same time. Colorimetric analytics of the reaction output without necessitating the opening of the reaction microchambers enables execution of the complete test workflow without any laboratory control that may otherwise be required stringently for safeguarding against carryover contamination. Seamless sample-to-answer workflow and machine learning-based readout further assures minimal human intervention for the test readout, thus eliminating inevitable inaccuracies stemming from erroneous execution of the test as well as subjectivity in interpreting the outcome. Our results further indicate the possibilities of upgrading the technology to predict the pathogenic load on the infected patients akin to the cyclic threshold value of the real-time PCR, when calibrated with reference to a wide range of ‘training’ data for the machine learner, thereby putting forward the same as viable alternative to the resource-intensive PCR tests that cannot be made readily accessible at underserved community settings.

Nanotechnology-based theranostic and prophylactic approaches against SARS-CoV-2.

SARS-CoV-2 (COVID-19) pandemic has been an unpredicted burden on global healthcare system by infecting over 700 million individuals, with approximately 6 million deaths worldwide. COVID-19 significantly impacted all sectors, but it very adversely affected the healthcare system. These effects were much more evident in the resource limited part of the world. Individuals with acute conditions were also severely impacted. Although classical COVID-19 diagnostics such as RT-PCR and rapid antibody testing have played a crucial role in reducing the spread of infection, these diagnostic techniques are associated with certain limitations. For instance, drawback of RT-PCR diagnostics is that due to degradation of viral RNA during shipping, it can give false negative results. Also, rapid antibody testing majorly depends on the phase of infection and cannot be performed on immune compromised individuals. These limitations in current diagnostic tools require the development of nanodiagnostic tools for early detection of COVID-19 infection. Therefore, the SARS-CoV-2 outbreak has necessitated the development of specific, responsive, accurate, rapid, low-cost, and simple-to-use diagnostic tools at point of care. In recent years, early detection has been a challenge for several health diseases that require prompt attention and treatment. Disease identification at an early stage, increased imaging of inner health issues, and ease of diagnostic processes have all been established using a new discipline of laboratory medicine called nanodiagnostics, even before symptoms have appeared. Nanodiagnostics refers to the application of nanoparticles (material with size equal to or less than 100nm) for medical diagnostic purposes. The special property of nanomaterials compared to their macroscopic counterparts is a lesser signal loss and an enhanced electromagnetic field. Nanosize of the detection material also enhances its sensitivity and increases the signal to noise ratio. Microchips, nanorobots, biosensors, nanoidentification of single-celled structures, and microelectromechanical systems are some of the most modern nanodiagnostics technologies now in development. Here, we have highlighted the important roles of nanotechnology in healthcare sector, with a detailed focus on the management of the COVID-19 pandemic. We outline the different types of nanotechnology-based diagnostic devices for SARS-CoV-2 and the possible applications of nanomaterials in COVID-19 treatment. We also discuss the utility of nanomaterials in formulating preventive strategies against SARS-CoV-2 including their use in manufacture of protective equipment, formulation of vaccines, and strategies for directly hindering viral infection. We further discuss the factors hindering the large-scale accessibility of nanotechnology-based healthcare applications and suggestions for overcoming them.

Accuracy of Initial Chest Computed Tomography Scan for Diagnosis of COVID-19: A Systematic Review

Abstract Background Coronavirus disease 2019 (COVID-19) is an infectious disease caused by severe acute respiratory syndrome coronavirus (SARS- COVID-19) with recent outbreak occurred on 11 March 2020. A computerized tomography (CT) scan for chest was proved to be a helpful tool for diagnosis of COVID-19 pneumonia. But little is known about its diagnostic accuracy. Aim of the Work This systematic review aims at determining current evidence regarding accuracy of initial chest CT scan for detection of COVID-19 infection. Patients and Methods We conducted an electronic search through different databases; PubMed, SCOPUS, Web of Science, Cochrane Central Register of Controlled Trials (CENTRAL). We included clinical trials whether randomized or nonrandomized prospective or retrospective cohorts, case control, cross sectional studies and case series larger than 10 cases). All studies that reported data about the diagnostic accuracy of initial chest CT for diagnosis of COVID-19 infection were enrolled. We evaluated the methodological quality for included studies using QUADAS-2 Risk of Bias tool. Results From a total 4617 screened citations, 40 studies met our inclusion criteria. We divided the identified studies into two groups: studies that reported both sensitivity and specificity (6 studies, n = 1431) and studies that included only positive COVID-19 cases from which we could estimate only sensitivity (34 studies, n = 3515). Among six studies that evaluated accuracy of CT chest, there was high sensitivity ranged from to 93% to 100% but CT chest specificity was low ranged from to 10% to 56%. Rest of 34 studies had a wide range of sensitivity between 61% and 100% with few studies reported sensitivity below 80%. Conclusion Our results suggest that CT chest is a great screening tool for COVID-19 with high sensitivity but it had low specificity. However, interpretation of CT chest along with other laboratory could enhance its diagnostic accuracy.

Enhancing COVID-19 Patients Detection using Deep Transfer Learning Technique Through X‑Ray Chest Images

This study addresses the urgent need for early detection of COVID-19 infection, considering its global impact. COVID-19 is caused by the severe acute respiratory syndrome Coronavirus 2 (SARS-CoV-2) and has affected more than 250 countries. Chest X-rays have been identified as a valuable tool for swift diagnosis of COVID-19 infection. In this research, we propose a composite approach to detect COVID-19 infection in its initial phases using radiographic images of the chest. To leverage existing models effectively, transfer learning is employed. Our model incorporates ensemble learning, combining transfer learning models like Efficient Net, Google Net, and XceptionNet. These models exhibit the ability to differentiate patients as COVID-19 positive, tuberculosis positive, pneumonia positive, or in good health. To evaluate the performance of our proposed model, we utilize two widely adopted datasets. Comparative analysis demonstrates that our technique surpasses current state-of-the-art models, as indicated by various performance measures.

Evaluation of the efficacy of tongue swab and saliva as samples for testing COVID-19 infection in symptomatic cases in comparison with nasopharyngeal swab.

Introduction. COVID-19 caused by SARS CoV-2 continues to be a major health concern globally. Methods for detection of the disease are necessary for public health efforts to monitor the spread of this disease as well as for detecting the emergence of new variants.Gap statement. Collection of Nasopharyngeal swab (NPS), the gold standard sample for the detection of COVID-19 infection by RT-qPCR is invasive and requires the expertise of a trained medical provider. This highlights the need for validating less invasive samples that can be self-collected without the need for trained medical provider.Aim. To validate saliva and tongue swab as potential samples for the diagnosis of COVID-19.Methodology. Adult and paediatric cases who had acute influenza like illness were enrolled in the study. The study involved comparison of Nucleic Acid Amplification Tests (NAAT) results for the detection of COVID-19 obtained by using saliva and tongue swab with that of NPS.Result and Conclusion. The sensitivity and specificity of saliva as sample for COVID-19 detection were found to be 71 and 88% respectively whereas those of tongue swab as sample were 78 and 90 %. Further validation was based on the positive and negative predictive values, the likelihood ratio, agreement percentage and the kappa statistic. The findings of the study point towards tongue swab and saliva as suitable alternative samples for the diagnosis of COVID-19 with a slightly higher accuracy and agreement for tongue swab than saliva. However considering the fatality of COVID-19, they are better suited for mass screening of people than for diagnosis.

Toward Data-Model-Agnostic Autonomous Machine-Generated Data Labeling and Annotation Platform: COVID-19 Autoannotation Use Case

Quick, early, and precise detection is important for diagnosis to control the spread of COVID-19 infection. Artificial Intelligence (AI) technology could certainly be used as a modulating tool to ease the detection, and help with the preventive steps further. Convolutional neural networks (CNNs) have achieved state-of-the-art performance in many visual recognition tasks. Nevertheless, most of these state-of-the-art networks highly rely on the availability of a high amount of labeled data, being an essential step in supervised machine learning tasks. Conventionally, this manual, mundane, and time-consuming process of annotating images is done by humans. Learning to localize or detect COVID-19 infection masks in our specific case study typically requires the collection of CT scan data that has been labeled with bounding boxes or similar annotations, which generally is limited. A technique that could perform such learning with much less annotations, and transfer the learned proposals that are algorithm-driven to generate more synthetic annotated samples would be helpful &amp; quite valuable. We present such a technique inspired by weakly trained mask region based convolutional neural networks (R-CNN) architecture for localization, in which the number of images with their pixel-level masks can be a small proportion of the total dataset, and then further improvise CNNs by inversely generating dense annotations on-the-go using an algorithmic-based computational approach. We focus on alleviating the bottleneck associated with deep learning models needing annotated data for training in an intuitive reverse engineering fashion through this work. Our proposed solution can certainly provide the prospect of automated labeling on-the-fly, thereby reducing much of the manual work. As a result, one can quickly train a precise COVID-19 infection detector with the leverage of autonomous frame-by-frame machine generated annotations. The model achieved mean precision accuracy (&#x0025;) of 0.99, 0.931, and 0.8 for train, validation, and test set, respectively. The results demonstrate that the proposed method can be adopted in a clinical setting for assisting radiologists, and also our fully autonomous approach can be generalized to any detection/recognition tasks at ease.

Challenges of deep learning diagnosis for COVID-19 from chest imaging

The COVID-19 pandemic has spread worldwide for over 2 years now. The pandemic raises a significant threat to global health due to its transmissibility and high pathogenicity. The current standard detection method for COVID-19, namely, reverse transcription–polymerase chain reaction (RT–PCR), is slow and inaccurate to help fight the pandemic. RT–PCR takes hours to days to report a single test result and has a high false-negative rate. As a result, an infected person with a negative test result may unknowingly continue to spread the virus. Thus, better detection methods are required to improve the control of COVID-19. With technology advancements in artificial intelligence and machine learning, deep-learning diagnostic studies to detect COVID-19 infection using medical chest imaging have emerged. In this paper, we review these studies by analyzing their approaches and highlighting their major challenges. These challenges include dataset cleanness, public dataset availability, capability to differentiate COVID-19 from unrelated viral pneumonia, and the difficulty in dealing with images from multiple points of view. Finally, we discuss various ideas and solutions to address the highlighted challenges in the reviewed papers.

Early Detection of COVID-19 in Female Athletes Using Wearable Technology.

Heart rate variability (HRV), respiratory rate (RR), and resting heart rate (RHR) are common variables measured by wrist-worn activity trackers to monitor health, fitness, and recovery in athletes. Variations in RR are observed in lower-respiratory infections, and preliminary data suggest changes in HRV and RR are linked to early detection of COVID-19 infection in nonathletes. Wearable technology measuring HRV, RR, RHR, and recovery will be successful for early detection of COVID-19 in NCAA Division I female athletes. Cohort study. Level 2. Female athletes wore WHOOP, Inc. bands through the 2020 to 2021 competitive season. Of the athletes who tested positive for COVID (n = 33), 14 had enough data to be assessed (N = 14; 20.0 ± 1.3 years; 69.8 ± 7.2 kg; 172.0 ± 8.3 cm). Roughly 2 weeks of noninfected days were used to set baseline levels of HRV, RR, recovery, and RHR to compare with -3, -2, and -1 days before a positive COVID-19 result. Increases in RR (P = 0.02) were detected on day -3. RHR (P < 0.01) and RR increased (P < 0.01), while HRV decreased (P < 0.05) on day -1, compared with baseline. Differences were noted in all variables on the day of the positive COVID-19 result: decreased HRV (P < 0.05) and recovery scores (P < 0.01), and increased RHR (P < 0.01) and RR (P < 0.01). In female athletes, wearable technology was successful in predicting COVID-19 infection through changes in RR 3 days before a positive test, and also HRV and RHR the day before a positive test. Wearable technology may be used, as part of a multifaceted approach, for the early detection of COVID-19 in elite athletes through monitoring of HRV, RR, and RHR for overall team health.

Modeling of COVID-19 in View of Rough Topology

Rough-based topology has become an important technique for decision making in numerous real-life problems. The main purpose of this research paper is to use some topological notions in the approximation space of a rough set. Firstly, some concepts of topological near open sets and rough concepts are presented. A new approximation structure based on the topological near open sets is introduced. We debate the properties of the new approximation structure. We included an algorithm for detecting COVID-19 infection by its side effects. We believe that our approach will be helpful for any future detection.

Comparison of pretrained transformer-based models for influenza and COVID-19 detection using social media text data in Saskatchewan, Canada.

The use of social media data provides an opportunity to complement traditional influenza and COVID-19 surveillance methods for the detection and control of outbreaks and informing public health interventions. The first aim of this study is to investigate the degree to which Twitter users disclose health experiences related to influenza and COVID-19 that could be indicative of recent plausible influenza cases or symptomatic COVID-19 infections. Second, we seek to use the Twitter datasets to train and evaluate the classification performance of Bidirectional Encoder Representations from Transformers (BERT) and variant language models in the context of influenza and COVID-19 infection detection. We constructed two Twitter datasets using a keyword-based filtering approach on English-language tweets collected from December 2016 to December 2022 in Saskatchewan, Canada. The influenza-related dataset comprised tweets filtered with influenza-related keywords from December 13, 2016, to March 17, 2018, while the COVID-19 dataset comprised tweets filtered with COVID-19 symptom-related keywords from January 1, 2020, to June 22, 2021. The Twitter datasets were cleaned, and each tweet was annotated by at least two annotators as to whether it suggested recent plausible influenza cases or symptomatic COVID-19 cases. We then assessed the classification performance of pre-trained transformer-based language models, including BERT-base, BERT-large, RoBERTa-base, RoBERT-large, BERTweet-base, BERTweet-covid-base, BERTweet-large, and COVID-Twitter-BERT (CT-BERT) models, on each dataset. To address the notable class imbalance, we experimented with both oversampling and undersampling methods. The influenza dataset had 1129 out of 6444 (17.5%) tweets annotated as suggesting recent plausible influenza cases. The COVID-19 dataset had 924 out of 11939 (7.7%) tweets annotated as inferring recent plausible COVID-19 cases. When compared against other language models on the COVID-19 dataset, CT-BERT performed the best, supporting the highest scores for recall (94.8%), F1(94.4%), and accuracy (94.6%). For the influenza dataset, BERTweet models exhibited better performance. Our results also showed that applying data balancing techniques such as oversampling or undersampling method did not lead to improved model performance. Utilizing domain-specific language models for monitoring users' health experiences related to influenza and COVID-19 on social media shows improved classification performance and has the potential to supplement real-time disease surveillance.

Learning without forgetting by leveraging transfer learning for detecting COVID-19 infection from CT images

COVID-19, a global pandemic, has killed thousands in the last three years. Pathogenic laboratory testing is the gold standard but has a high false-negative rate, making alternate diagnostic procedures necessary to fight against it. Computer Tomography (CT) scans help diagnose and monitor COVID-19, especially in severe cases. But, visual inspection of CT images takes time and effort. In this study, we employ Convolution Neural Network (CNN) to detect coronavirus infection from CT images. The proposed study utilized transfer learning on the three pre-trained deep CNN models, namely VGG-16, ResNet, and wide ResNet, to diagnose and detect COVID-19 infection from the CT images. However, when the pre-trained models are retrained, the model suffers the generalization capability to categorize the data in the original datasets. The novel aspect of this work is the integration of deep CNN architectures with Learning without Forgetting (LwF) to enhance the model’s generalization capabilities on both trained and new data samples. The LwF makes the network use its learning capabilities in training on the new dataset while preserving the original competencies. The deep CNN models with the LwF model are evaluated on original images and CT scans of individuals infected with Delta-variant of the SARS-CoV-2 virus. The experimental results show that of the three fine-tuned CNN models with the LwF method, the wide ResNet model’s performance is superior and effective in classifying original and delta-variant datasets with an accuracy of 93.08% and 92.32%, respectively.