Centralized CNN-GRU Model by Federated Learning for COVID-19 Prediction in India

In 2019, the corona virus was found in Wuhan, China. The corona virus has traveled several countries in the world from the beginning of 2020. The early estimation of COVID-19 cases is one of the efficient approaches to control the pandemic. Many researchers had proposed the deep learning model for the efficient estimation of COVID-19 cases for different provinces in the world. The research work had not focused on the discussion of robustness in the model. In this study, centralized federated-convolutional neural network–gated recurrent unit (Fed-CNN–GRU) model is proposed for the estimation of active cases per day in different provinces of India. In India, the uneven transmission of COVID-19 virus was seen in 36 provinces due to the different geographical areas and population densities. So, the methodology of this study had focused on the development of single deep learning algorithm, which is robust and reliable to estimate the active cases of COVID-19 in different provinces of India. The concept of transfer and federated learning is involved to enhance the estimation of active cases of COVID-19 by the CNN–GRU model. The study had considered the active cases per day dataset for 36 provinces in India from 12 March, 2020 to 17 January, 2022. Based on the study, it is proven that the centralized CNN–GRU model by federated learning had captured the transmission dynamics of COVID-19 in different provinces with an enhanced result. IEEE

Serological Immune Response Following ChAdOx1 nCoV-19 Vaccine (Covishield®) in Patients with Liver Cirrhosis.

INTRODUCTION: Data are limited on antibody response to the ChAdOx1 nCoV-19 vaccine (AZD1222; Covishield®) in cirrhosis. We studied the antibody response following two doses of the ChAdOx1 vaccine, given 4-12 weeks apart, in cirrhosis. METHODS: Prospectively enrolled, 131 participants (71% males; age 50 (43-58); alcohol-related etiology 14, hepatitis B 33, hepatitis C 46, cryptogenic 21, autoimmune 9, others 8; Child-Turcott-Pugh class A/B/C 52/63/16). According to dose intervals, the participants were grouped as ≤6 weeks (group I), 7-12 weeks (group II), and 13-36 weeks (group III). Blood specimens collected at ≥4 weeks after the second dose were tested for anti-spike antibody titre (ASAb; positive ≥ 0.80 U/mL) and neutralizing antibody (NAb; positive ≥20% neutralization) using Elecsys Anti-SARS-CoV-2 S (Roche) and SARS-CoV-2 NAb ELISA Kit (Invitrogen), respectively. Data are expressed as number (proportion) and median (interquartile range) and compared using non-parametric tests. RESULTS: Overall, 99.2% and 84% patients developed ASAb (titre 5440 (1719-9980 U/mL)) and NAb (92 (49.1-97.6%)), respectively. When comparing between the study groups, the ASAb titres were significantly higher in group II than in group I (2613 (310-7518) versus 6365 (2968-9463), p = 0.027) but were comparable between group II and III (6365 (2968-9463) versus 5267 (1739-11,653), p = 0.999). Similarly, NAb was higher in group II than in group I (95.5 (57.6-98.0) versus 45.9 (15.4-92.0); p < 0.001), but not between the groups II and III (95.5 (57.6-98.0) versus 92.4 (73.8-97.5); p = 0.386). CONCLUSION: Covishield® induces high titres of ASAb and NAb in cirrhosis. A higher titre is achieved if two doses are given at an interval of more than six weeks.

Antibody Response to ChAdOx1 nCoV-19 (AZD1222) Vaccine in Kidney Transplant Recipients.

Kidney transplant recipients (KTRs) are at a much higher risk of complications and death following COVID-19 and are poor vaccine responders. The data are limited on the immune response to Covishield® in KTRs. We prospectively recruited a cohort of 67 KTRs aged >18 between April 2021 and December 2021. Each participant was given two intramuscular doses of Covishield®, each of 0.5 mL, at an interval of 12 weeks. A blood specimen of 5.0 mL was collected from each participant at two points within a few days before administering the first dose of the vaccine and at any time between 4-12 weeks after administering the second dose. The sera were tested for anti-RBD antibody (ARAb) titre and neutralising antibody (NAb). An ACE2 competition assay was used as a proxy for virus neutralization. According to the prior COVID-19 infection, participants were grouped as (i) group A: prior symptomatic COVID-19 infection, (ii) group B: prior asymptomatic COVID-19 infection as evidenced by detectable ARAb in the prevaccination specimen, (iii) Group C: no prior infection with COVID-19, (iv) group D: Unclassified, i.e., participants had no symptoms suggestive of COVID-19, but their prevaccination specimen was not available for ARAb testing before vaccination. Fifty of sixty-seven participants (74.6%) provided paired specimens (group A 14, group B 27, and group C 9) and 17 participants (25.4%) provided only postvaccination specimens (group D). In the overall cohort (n = 67), 91% and 77.6% of participants developed ARAb and NAb, respectively. Their ARAb titre and NAb proportion were 2927 (520-7124) U/mL and 87.9 (24.4-93.2) %, respectively. Their median ARAb titre increased 65.6 folds, from 38.2 U/mL to 3137 U/mL. Similarly, the proportion of participants with NAb increased from 56% to 86%, and the NAb proportion raised 2.7 folds, from 23% to 91%. A comparison of vaccine response between the study groups showed that all those with or without prior COVID-19 infection showed a significant rise in ARAb titre (p < 0.05) and NAb proportion (p < 0.05) after the two doses of vaccine administration. The median value of folds rise in anti-RBD and NAb between groups A and B were comparable. Hence, ARAb is present in more than 3/4th of KTRs before the ChAdOx1 vaccine in India. The titer of ARAb and the proportion of NAb significantly increased after the two doses of the ChAdOx1 vaccine in KTRs.

Unsupervised Deep learning-based Feature Fusion Approach for Detection and Analysis of COVID-19 using X-ray and CT Images

Aims: This study investigates an unsupervised deep learning-based feature fusion approach for the detection and analysis of COVID-19 using chest X-ray (CXR) and Computed tomography (CT) images. Background: The outbreak of COVID-19 has affected millions of people all around the world and the disease is diagnosed by the reverse transcription-polymerase chain reaction (RT-PCR) test which suffers from a lower viral load, and sampling error, etc. Computed tomography (CT) and chest X-ray (CXR) scans can be examined as most infected people suffer from lungs infection. Both CT and CXR imaging techniques are useful for the COVID-19 diagnosis at an early stage and it is an alternative to the RT-PCR test. Objective: The manual diagnosis of CT scans and CXR images are labour-intensive and consumes a lot of time. To handle this situation, many AI-based solutions are researched including deep learning-based detection models, which can be used to help the radiologist to make a better diagnosis. However, the availability of annotated data for COVID-19 detection is limited due to the need for domain expertise and expensive annotation cost. Also, most existing state-of-the-art deep learning-based detection models follow a supervised learning approach. Therefore, in this work, we have explored various unsupervised learning models for COVID-19 detection which does not need a labelled dataset. Methods: In this work, we propose an unsupervised deep learning-based COVID-19 detection approach that incorporates the feature fusion method for performance enhancement. Four different sets of experiments are run on both CT and CXR scan datasets where convolutional autoencoders, pre-trained CNNs, hybrid, and PCA-based models are used for feature extraction and K-means and GMM techniques are used for clustering. Results: The maximum accuracy of 84% is achieved by the model Autoencoder3-ResNet50 (GMM) on the CT dataset and for the CXR dataset, both Autoencoder1-VGG16 (KMeans and GMM) models achieved 70% accuracy. Conclusion: Our proposed deep unsupervised learning, feature fusion-based COVID-19 detection approach achieved promising results on both datasets. It also outperforms four well-known existing unsupervised approaches. © 2022 Ravi and Pham.

Elevation of neural injury markers in patients with neurologic sequelae after hospitalization for SARS-CoV-2 infection.

Patients with SARS-CoV-2 infection (COVID-19) risk developing long-term neurologic symptoms after infection. Here, we identify biomarkers associated with neurologic sequelae one year after hospitalization for SARS-CoV-2 infection. SARS-CoV-2-positive patients were followed using post-SARS-CoV-2 online questionnaires and virtual visits. Hospitalized adults from the pre-SARS-CoV-2 era served as historical controls. 40% of hospitalized patients develop neurological sequelae in the year after recovery from acute COVID-19 infection. Age, disease severity, and COVID-19 infection itself was associated with additional risk for neurological sequelae in our cohorts. Glial fibrillary astrocytic protein (GFAP) and neurofilament light chain (NF-L) were significantly elevated in SARS-CoV-2 infection. After adjusting for age, sex, and disease severity, GFAP and NF-L remained significantly associated with longer term neurological symptoms in patients with SARS-CoV-2 infection. GFAP and NF-L warrant exploration as biomarkers for long-term neurologic complications after SARS-CoV-2 infection.

Basic virology of SARS-CoV 2

Background Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), is a highly transmissible pathogenic coronavirus emerged in late 2019 causing a pandemic of acute respiratory disease, named ‘coronavirus disease 2019’ (COVID-19). It has spread fast all over the world posing an extraordinary threat to global public health. Along with SARS-CoV-2, there are seven human coronaviruses. Those causing mild diseases are the 229E, OC43, NL63 and HKU1, and the pathogenic ones are SARSCoV, MERS-CoV and SARS-CoV-2. Objective This review has highlighted the basic virology of SARS CoV-2 including its origin, structure, genomic characteristics, pathogenesis, immunological response and clinical manifestation along with the key difference of SARS CoV2 from the previous Coronaviruses. Content Coronaviruses are spherical and enveloped with club-shaped spikes on the surface. It has a large positive sense, single stranded RNA genome within the nucleocapsid with a helical symmetry. It has been known to cause infection to innumerable mammalian hosts, like humans, cats, bats, civets, dogs, and camels. The viral genome contains four major structural proteins: the spike (S), membrane (M), envelope (E) and the nucleocapsid (N) protein encoded within the 3’ end of the genome. Virus binds to the host cell by the S protein with specific receptor. Following receptor binding, the virus enters host cell cytosol and there is fusion of the viral and cellular membranes followed by the translation of the viral genomic RNA. Following the viral replication and sub-genomic RNA synthesis, there is formation of the mature virus. The virions are then transported to the cell surface in vesicles and are released by exocytosis.

Analysis of Subfeature for Classification in Data Mining

Feature selection is a crucial part of data mining applications in the diversity area of data analysis. Most of the feature selection methods issue features analysis for classification, which is not sufficient for classification by feature selection techniques. Classification is involved in traditional classes with redundant feature data, which is not helpful in the database for further analysis. There is a more negligible effect on feature analysis by taking an existing class. So, it proposed to obtain a new class or subclass using the analysis of feature's tiny data named subfeature (SF) data of corresponding instances from a traditional class. These data are involved in the limited number of significant instances where it generates the new class. It is a challenging task to find such instances with subfeatures from any database. Thus, this article proposed the optimization model based on the Lagrangian multiplier to find such data and analyze for a new class from a traditional class. Several algorithms have been taken to explain subfeature data. The theoretical approaches such as domain- and variance-based subfeatures and the convergence of subfeature are used to select subfeatures with the effectiveness of the proposed model. Furthermore, several classifiers with searching methods and statistical methods (i.e., local and global variances) analyze the classification through subfeature data. Result analysis in the experiment on different datasets shows that the proposed model is beneficial toward novel classes based on selected subfeature data.

Lung-GANs: Unsupervised Representation Learning for Lung Disease Classification Using Chest CT and X-Ray Images

Lung diseases are a tremendous challenge to the health and life of people globally, accounting for 5 out of 30 most common causes of death. Early diagnosis is crucial to help in faster recovery and improve long-term survival rates. Deep learning techniques offer a great promise for automated, fast, and reliable detection of lung diseases from medical images. Specifically, convolutional neural networks have accomplished encouraging results in disease detection. In spite of that, the performance of such supervised models depends heavily on the availability of large labeled data, the collection of which is an expensive and tedious task, specially for a novel disease. Therefore, in this article, we propose a deep unsupervised framework to classify lung diseases from chest CT and X-ray images. Our framework introduces multiple-layer generative adversarial networks called Lung-GANs that learn interpretable representations of lung disease images using only unlabeled data. We use the lung features learned by the model to train a support vector machine and a stacking classifier. We demonstrate through experiments that the proposed method outperforms the current state-of-the-art unsupervised models in lung disease classification. Our model obtained an accuracy of 94x0025;x2013;99.5x0025;on all the six large-scale publicly available lung disease datasets used in this study. Hence, the proposed framework will simplify lung disease detection by reducing the time for diagnosis and increasing the convenience of diagnostics. IEEE

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 And 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 (%) 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. IEEE

Real-world patient-reported and clinical outcomes of BNT162b2 mRNA COVID-19 vaccine in patients with cancer

Background: Most COVID-19 (C19) vaccine trials excluded patients with active cancer. Here, we report our real-world patient-reported and clinical outcomes of BNT162b2 mRNA C19 vaccine in patients with cancer. Methods: Our institutional Data-Driven Determinants for COVID-19 Oncology Discovery Effort (D3CODE) follows a longitudinal observational cohort of pts w cancer getting C19 vaccine. Pts complete a validated PRO tool, MD Anderson Symptom Inventory (MDASI, 13 core, 6 interference plus 17 items of symptoms from prior vaccine trials) pre-dose 1, then daily x 6d, then weekly, then on day of dose 2, then daily x 6d, then weekly x 3w. Demographics, cancer variables, prior immune checkpoint inhibitors (ICI), C19 status pre- & post-vaccine are aggregated via Syntropy platform: Palantir Foundry. Primary outcome is incidence of PRO symptoms bw dose 1 & 2 across AYA 15-39y, mid-age 40-64y & senior 65y+ cohorts. Secondary outcomes include PRO symptom incidence post-dose 2, post-vaccine change in cancer symptoms, post-vaccine symptom severity based on prior ICI, and confirmed C19 > 7 days post-dose 2. First planned 8-wk interim analysis is reported here. Results: 6388 pts w cancer (4973 w mets) received a BNT162b2 vaccine dose (4811 both doses, 1577 received one & await dose 2). Overall, median age 64y (range 16-95y);382 AYAs, 2927 mid-age, 3079 seniors (65-70y n = 1158, 70-79y n = 1521, 80-89y n = 378, 90y+ n = 22). 4099 (64%) are White, 823 (13%) AA, 791 (12%) Hispanic, 441 (7%) Asians. Primary cancers: breast (1397), GU (821), heme (775), thoracic/HN (745), and CRC (385). Prior to dose 1, 1862 had no prior systemic tx while 4526 pts did including 3243 who had only non-IO tx (chemo, targeted tx), 1,283 had immunotherapy including 857 who had ICIs prior to dose 1. Patient-reported symptoms after C19 Vaccine: Of 6388 pts, 4714 (74% response rate, median age 67y, range 16-95y) completed 16485 PRO surveys. After 2 doses, seniors reported lower mean scores vs mid-age or AYAs on 22 of 36 symptoms including injection site pain, palpitations, itch, rash, malaise, fevers/chills, arthralgia, myalgia, headache, pain, fatigue, nausea, disturbed sleep, distress (p < 0.05). Pts w prior ICIs had higher severity of itch, rash (p < 0.05) from baseline after both dose 1 & 2 vs pts without systemic tx. Post dose 1, pts with prior ICI had higher increase in fatigue, malaise, itch, rash, myalgia, anorexia from their baseline vs pts without systemic tx (p < 0.05). C19 Outcomes: Of 6388 pts, 616 had a C19 test at any time post-dose 1: 23 (0.36%) tested positive of whom 20 (0.3%) were between dose 1 & 2;two (0.031%) were within 7 days post-dose 2, and one patient (0.016%) tested positive 16 days after dose 2, requiring admission. Conclusions: This real-world observational cohort demonstrates post-vaccine symptom burden and outcomes in patients with cancer. Second interim analysis is planned at 16 weeks.

Development and validation of a novel patient-reported outcomes (PRO) measure for symptom burden in patients with cancer and COVID-19 infection

Background: The symptom burden experienced by patients with cancer who contract the COVID-19 (C19) infection remains to be fully understood. To accurately assess this symptom burden, we developed a valid, reliable patient-reported outcome (PRO) measure of C19 symptoms combined with a known measure of cancer symptom burden. Methods: Within the institutional initiative on COVID-19 and cancer named Data- Driven Determinants for COVID-19 Oncology Discovery Effort (D3CODE), patients with cancer and PCR-positive C19 tests were invited to participate in this longitudinal study after providing consent. Pts completed the EQ-5D-5L and the 13 symptom severity and 6 interference items of the core MD Anderson Symptom Inventory (MDASI) plus 14 COVID-specific symptom items generated from literature and expert review. Items were measured on a 0-10 scale, 0 = none to 10 = worst imaginable symptom or interference. Demographic and disease information was collected. Psychometric procedures determined validity and reliability of the MDASI-COVID. Results: 600 pts enrolled, mean age 56.5y (range 20 to 91y). 59% female, 80% white. 78% solid tumors, 19% heme cancers. 12.5% required hospitalization for C19. Median number of days between positive C19 test and PRO completion was 17 days. Mean overall health rating on EQ-5D-5L was 78.3 (SD 19.6), best being 100. Highest mean (M) severity symptoms on the MDASI-COVID were fatigue (M 3.45, SD 2.17), drowsiness (M 2.50, SD 2.89), sleep disturbance (M 2.44, SD 2.99), malaise (M 2.37, SD 3.05), and distress (M 2.27, SD 2.90). Most severe (≥ 7) symptoms) reported were fatigue (21.3% of pts), change in taste (14.8%), change in smell (14.4%), malaise (14.3%), sleep disturbance (14.3%), and drowsiness (14%). Internal consistency (Cronbach α) of the 27 symptom items was 0.957, of the 6 interference items was 0.937. Mean severity of the 27 symptom items was significantly correlated with overall EQ-5D-5L health rating (correlation = -0.45, P < 0.0005), demonstrating concurrent validity. Mean symptom severity and interference showed known-group validity between patients who required C19 hospitalization (symptom M 2.32, SD 2.09;interference M 3.29, SD 3.02) and those who did not (symptom M 1.69, SD 1.85;interference M 2.20, SD 2.64) (symptom P 0.007;interference P 0.004). Conclusions: We have validated a novel PRO, the MDASI-COVID, to quantify the combined symptom burden in patients with cancer and COVID-19. This measure allows longitudinal evaluation of COVID-19 on cancer symptom burden and provide clinicians with an accurate tool for ongoing symptom assessment and management. Longitudinal analysis on long-term symptoms related to COVID-19 and cancer are ongoing.