Evaluation of the impact of COVID-19 pandemic on hospital admission related to common infections (preprint)

Background: Antimicrobial resistance (AMR) is a multifaceted global challenge, partly driven by inappropriate antibiotic prescribing. The COVID-19 pandemic impacted antibiotic prescribing for common bacterial infections. This highlights the need to examine risk of hospital admissions related to common infections, excluding COVID-19 infections during the pandemic. Methods: With the approval of NHS England, we accessed electronic health records from The Phoenix Partnership (TPP) through OpenSAFELY platform. We included patients with primary care diagnosis of common infections, including lower respiratory tract infection (LRTI), upper respiratory tract infections (URTI), and lower urinary tract infection (UTI), from January 2019 to August 2022. We excluded patients with a COVID-19 record 90 days before to 30 days after the infection diagnosis. Using Cox proportional-hazard regression models, we predicted risk of infection-related hospital admission in 30 days follow-up period after the diagnosis. Results: We found 12,745,165 infection diagnoses from January 2019 to August 2022. Of them, 80,395 (2.05%) cases were admitted to hospital in the follow-up period. Counts of hospital admission for infections dropped during COVID-19, e.g., LRTI from 3,950 in December 2019 to 520 in April 2020. Comparing those prescribed an antibiotic to those without, reduction in risk of hospital admission were largest with LRTI (adjusted odds ratio (OR) of 0.35; 95% CI, 0.35-0.36) and UTI (adjusted OR 0.45; 95% CI, 0.44-0.46), compared to URTI (adjusted OR 1.04; 95% CI, 1.03-1.06). Conclusion: Large effectiveness of antibiotics in preventing complications related to LRTI and UTI can support better targeting of antibiotics to patients with higher complication risks.

Targeting Multiple Conserved T-Cell Epitopes for Protection against COVID-19 Moderate-Severe Disease by a Pan-Sarbecovirus Vaccine (preprint)

Background Most of current approved vaccines, based on a Spike-only as single immunogen, fall short of producing a full-blown T-cell immunity. SARS-CoV-2 continues to evolve with ever-emergent higher-contagious mutants that may take a turn going beyond Omicron to bring about a new pandemic outbreak. New recombinant SARS-CoV-2 species could be man-made through genetic manipulation to infect systemically. Development of composition-innovated, pan-variant COVID-19 vaccines to prevent from hospitalization and severe disease, and to forestall the next pandemic catastrophe, is an urgent global objective. Methods and findings In a retrospective, e-questionnaire Observational Study, extended from a clinical Phase-2 trial conducted in Taiwan, during the prime time of Omicron outbreak dominated by BA.2 and BA.5 variants, we investigated the preventive effects against COVID-19 moderate-severe disease (hospitalization and ICU admission) by a pan-Sarbecovirus vaccine UB-612 that targets monomeric S1-RBD-focused subunit protein and five designer peptides comprising sequence-conserved, non-mutable helper and cytotoxic T lymphocyte (Th/CTL) epitopes derived from Spike (S2), Membrane (M) and Nucleocapsid (N) proteins. Per UB-612 vaccination, there were no hospitalization and ICU admission cases (0% rate, 6 months after Omicron outbreak) reported [≥]14 months post-2nd dose of primary series, and [≥]10 months post-booster (3rd dose), to which the potent memory cytotoxic CD8 T cell immunity may be the pivotal in control of the infection disease severity. Six months post-booster, the infection rate (asymptomatic and symptomatic mild) was only 1.2%, which increased to 27.8% observed [≥]10 months post-booster. The notable protection effects are in good alignment with a preliminary Phase-3 heterologous booster trial report showing that UB-612 can serve as a competent booster substitute for other EUA-approved vaccine platforms to enhance their seroconversion rate and viral-neutralizing titer against Omicron BA.5. Conclusions UB-612, a universal multitope vaccine promoting full-blown T cell immunity, may work as a competent primer and booster for persons vulnerable to Sarbecovirus infection. Trial Registration. Clinical Trials.gov ID: NCT04773067.

Feasibility Study on Browser-Based Federated Machine Learning (FML) Architecture for Medical Application

Traditional approaches to Artificial Intelligence (AI) based medical image classification requires huge amounts of data sets to be stored in a centralized server for analysis and training. In medical applications, data privacy and ownership may pose a challenge. In addition, costs incurred by data transfer and cloud server may pose a challenge to implementing a large dataset. This work studies the feasibility of a decentralized, browser-based Artificial Intelligence (AI) federated machine learning (FML) architecture. The proposed work studies the feasibility of bringing training and inference to the browser, hence removing the need to transfer raw data to a centralized server. If feasible, the system allows practitioners to compress and upload their pre-trained model to the server instead of raw data. This allows medical practitioners to update the model without the need to reveal their raw data. A sandbox system was implemented by applying transfer learning on MobileNet V3 and was tested with chest X-ray image datasets from COVID-19, viral pneumonia, and normal patients to simulate medical usage environment. The training speed, model performance and inference speed were tested on a PC browser and mobile phone with various levels of network throttling and image degradation. © 2022 IEEE.

An Analysis of Public Service Satisfaction of Tourists at Scenic Spots: The Case of Xiamen City

The outbreak of the COVID-19 epidemic has greatly hindered the development of the tourism industry. It is urgent for the city to improve the quality of public service in scenic spots, so as to attract more tourists and achieve sustainable development. With a literature review and reference to some guidance, the evaluation indicator system of public service satisfaction of scenic spots was constructed based on the analytic hierarchy process. Then, we distributed two questionnaires to complete the study. The first is the expert questionnaire for the evaluation indicator system. We used YAAHP software to process the questionnaire data and calculated the weight of each indicator, which provided a basis for the following analysis. The second is the questionnaire distributed to tourists of Xiamen. Then, we used the fuzzy comprehensive evaluation method to analyze the satisfaction of public services in scenic spots. The outcomes show that tourists are overall satisfied with the public services of scenic spots in Xiamen. However, there are still some problems, such as lagging information service, inadequate security, and supervision in the scenic spot. Therefore, the city and scenic spots should improve the level of smart tourism service, strengthen the construction of hardware and software facilities, and focus on the protection of tourists' rights and interests.

UB-612 Multitope Vaccine Targeting SARS-CoV-2 Spike and Non-Spike Proteins Provides Broad and Durable Immune Responses (preprint)

IRBThe SARS-CoV-2 non-Spike (S) structural protein targets of nucleocapsid (N), membrane (M) and envelope (E), critical in the host cell interferon response and memory T-cell immunity, have been grossly overlooked since the inception of COVID vaccine development. To pursue a universal (pan-sarbecovirus) vaccine against ever-emergent future mutants, we explored booster immunogenicity of UB-612, a multitope-vaccine that contains S1-RBD-sFc protein and sequence-conserved rationally designed promiscuous Th and CTL epitope peptides on the Sarbecovirus N, M and S2 proteins. To a subpopulation of infection-free participants (aged 18-85 years) involved in a two-dose Phase-2 trial, a UB-612 booster (third dose) was administered 6-8 months after the second dose. The immunogenicity was evaluated at 14 days post-booster with overall safety monitored until the end of study. The booster induced high viral-neutralizing antibodies against live Wuhan WT (VNT50, 1,711) and Delta (VNT50, 1,282); and against pseudovirus WT (pVNT50, 11,167) vs. Omicron BA.1/BA.2/BA.5 variants (pVNT50, 2,314/1,890/854), respectively. The lower primary neutralizing antibodies in the elderly were uplifted upon boosting to approximately the same high level in young adults. UB-612 also induced potent, durable Th1-oriented (IFN-{gamma}+-) responses (peak/pre-boost/post-boost SFU/106 PBMCs, 374/261/444) along with robust presence of cytotoxic CD8+ T cells (peak/pre-boost/post-boost CD107a+-Granzyme B+, 3.6%/1.8%/1.8%). Booster vaccination is safe and well tolerated without SAEs. By recognition against epitopes on Spike (S1-RBD and S2) and non-Spike (N and M) structure proteins, UB-612 provides potent, broad and long-lasting B-cell and T-cell memory immunity and offers a potential as a universal vaccine to fend off Omicrons and new VoCs.

UB-612, a Multitope Universal Vaccine Eliciting a Balanced B and T Cell Immunity against SARS-CoV-2 Variants of Concern (preprint)

Importance. The SARS-CoV-2 non-spike structural proteins of nucleocapsid (N), membrane (M) and envelope (E) are critical in the host cell interferon response and memory T-cell immunity and have been grossly overlooked in the development of COVID vaccines. Objective. To determine the safety and immunogenicity of UB-612, a multitope vaccine containing S1-RBD-sFc protein and rationally-designed promiscuous peptides representing sequence-conserved Th and CTL epitopes on the Sarbecovirus nucleocapsid (N), membrane (M) and spike (S2) proteins. Design, setting and participants. UB-612 booster vaccination was conducted in Taiwan. A UB-612 booster dose was administered 6-8 months post-2nd dose in 1,478 vaccinees from 3,844 healthy participants (aged 18-85 years) who completed a prior placebo (saline)-controlled, randomized, observer-blind, multi-center Phase-2 primary 2-dose series (100-ug per dose; 28-day apart) of UB-612. The interim safety and immunogenicity were evaluated until 14 days post-booster. Exposure. Vaccination with a booster 3rd-dose (100-ug) of UB-612 vaccine. Main outcomes and measures. Solicited local and systemic AEs were recorded for seven days in the e-diaries of study participants, while skin allergic reactions were recorded for fourteen days. The primary immunogenicity endpoints included viral-neutralizing antibodies against live SARS-CoV-2 wild-type (WT, Wuhan strain) and live Delta variant (VNT50), and against pseudovirus WT and Omicron variant (pVNT50). The secondary immunogenicity endpoints included anti-S1-RBD IgG antibody, S1-RBD:ACE2 binding inhibition, and T-cell responses by ELISpot and Intracellular Staining. Results. No post-booster vaccine-related serious adverse events were recorded. The most common solicited adverse events were injection site pain and fatigue, mostly mild and transient. The UB-612 booster prompted a striking upsurge of neutralizing antibodies against live WT Wuhan strain (VNT50, 1,711) associated with unusually high cross-neutralization against Delta variant (VNT50, 1,282); and similarly with a strong effect against pseudovirus WT (pVNT50, 6,245) and Omicron variant (pVNT50, 1,196). Upon boosting, the lower VNT50 and pVNT50 titers of the elderly in the primary series were uplifted to the same levels as those of the young adults. The UB-612 also induced robust, durable VoC antigen-specific Th1-oriented (IFN-{gamma}+-) responses along with CD8+ T-cell (CD107a+-Granzyme B+) cytotoxicity. Conclusions and relevance. With a pronounced cross-reactive booster effect on B- and T-cell immunity, UB-612 may serve as a universal vaccine booster for comprehensive immunity enhancement against emergent VoCs. Trial registration. [ClinicalTrials.gov: NCT04773067]

Subtyping of major SARS-CoV-2 variants reveals different transmission dynamics (preprint)

SARS-CoV-2 continues to evolve, causing waves of the pandemic. Up to March 2022, eight million genome sequences have accumulated, which are classified into five major variants of concern. With the growing number of sequenced genomes, analysis of the big dataset has become increasingly challenging. Here we developed systematic approaches for comprehensive subtyping and pattern recognition for transmission dynamics. By analyzing the first two million viral genomes as of July 2021, we found that different subtypes of the same variant exhibited distinct temporal trajectories. For example, some Delta subtypes did not spread rapidly, while others did. We identified sets of characteristic single nucleotide variations (SNVs) that appeared to enhance transmission or decrease efficacy of antibodies for some subtypes of the Delta and Alpha variants. We also identified a set of SNVs that appeared to suppress transmission or increase viral sensitivity to antibodies. These findings are later confirmed in an analysis of six million genomes as of December 2021. For the Omicron variant, the dominant type in the world, we identified the subtypes with enhanced and suppressed transmission in an analysis of seven million genomes as of January 2022 and further confirmed the findings in a later analysis of eight million genomes as of March 2022. While the "enhancer" SNVs exhibited an enriched presence on the spike protein, the suppressor SNVs are mainly elsewhere. Disruption of the SNV correlation largely destroyed the enhancer-suppressor phenomena. These results suggest the importance of fine subtyping of variants, and point to potential complex interactions among SNVs.

Potent Anti-Delta Effect By a Booster Third-Dose of UB-612, a Precision-Designed SARS-CoV-2 Multitope Protein-Peptide Vaccine (preprint)

SARS-CoV-2 breakthrough infection occurs due to waning immunity time-to-vaccine, to which the globally-dominant, highly-contagious Delta variant is behind the scene. In the primary 2-dose and booster series of clinical Phase-1 trial, UB-612 vaccine, which contains S1-RBD and synthetic Th/CTL peptide pool for activation of humoral and T-cell immunity, induces substantial, prolonged viral-neutralizing antibodies that goes parallel with a long-lasting T-cell immunity; and a booster (3rd ) dose can prompt recall of memory immunity to induce profound, striking antibodies with the highest level of 50% viral-neutralizing GMT titers against live Delta variant reported for any vaccine. The unique design of S1-RBD only plus multitope T-cell peptides may have underpinned UB-612’s potent anti-Delta effect, while the other full S protein-based vaccines are affected additionally by mutations in the N-terminal domain sequence which contains additional neutralizing epitopes. UB-612, safe and well-tolerated, could be effective for boosting other vaccine platforms that have shown modest homologous boosting. [Funded by United Biomedical Inc., Asia; ClinicalTrials.gov ID: NCT04967742 and NCT04545749]

Broad neutralizing nanobody against SARS-CoV-2 engineered from pre-designed synthetic library (preprint)

SARS-CoV-2 infection is initiated with Spike glycoprotein binding to the receptor of human angiotensin converting enzyme 2 via its receptor binding domain. Blocking this interaction is considered as an effective approach to inhibit virus infection. Here we report the discovery of a neutralizing nanobody, VHH60, directly produced from a humanized synthetic nanobody library. VHH60 competes with human ACE2 to bind the receptor binding domain of the Spike protein with a KD of 2.56 nM, inhibits infections of both live SARS-CoV-2 and pseudotyped viruses harboring wildtype, escape mutations and prevailing variants at nanomolar level. VHH60 also suppresses SARS-CoV-2 infection and propagation 50-fold better and protects mice from death two times longer than that of control group after live virus inoculation on mice. VHH60 therefore is a powerful synthetic nanobody with a promising profile for disease control against COVID19.

Blockchain-Federated-Learning and Deep Learning Models for COVID-19 Detection Using CT Imaging.

With the increase of COVID-19 cases worldwide, an effective way is required to diagnose COVID-19 patients. The primary problem in diagnosing COVID-19 patients is the shortage and reliability of testing kits, due to the quick spread of the virus, medical practitioners are facing difficulty in identifying the positive cases. The second real-world problem is to share the data among the hospitals globally while keeping in view the privacy concerns of the organizations. Building a collaborative model and preserving privacy are the major concerns for training a global deep learning model. This paper proposes a framework that collects a small amount of data from different sources (various hospitals) and trains a global deep learning model using blockchain-based federated learning. Blockchain technology authenticates the data and federated learning trains the model globally while preserving the privacy of the organization. First, we propose a data normalization technique that deals with the heterogeneity of data as the data is gathered from different hospitals having different kinds of Computed Tomography (CT) scanners. Secondly, we use Capsule Network-based segmentation and classification to detect COVID-19 patients. Thirdly, we design a method that can collaboratively train a global model using blockchain technology with federated learning while preserving privacy. Additionally, we collected real-life COVID-19 patients' data open to the research community. The proposed framework can utilize up-to-date data which improves the recognition of CT images. Finally, we conducted comprehensive experiments to validate the proposed method. Our results demonstrate better performance for detecting COVID-19 patients.

Safety Monitoring of COVID Vaccines and Perception of Post-vaccination Side-Effects: Preliminary Findings From The First Month of Routine Monitoring in a Hospital Vaccination Setting of China (preprint)

Objectives: This study aims to explore the occurrence of post-vaccination side-effects from COVID-19 vaccines and its affecting factors in a hospital vaccination setting of China.Results: A total of 811 vaccinees aged 17 to 58 years, who finished the full package of two doses in February 2021, have been recruited at the second vaccination uptake. Among all, there have been 66 participants who reported one or more mild side effects, while none of them developed severe cases. Those with history of immune deficiency were more likely to report side effect(s). Although with several concerns, most participants showed willingness to get vaccinated (98.8%) with relevant high proportions of perceived safety (99.5%) and effectiveness (97.3%).

Blockchain based Privacy-Preserved Federated Learning for Medical Images: A Case Study of COVID-19 CT Scans (preprint)

Medical health care centers are envisioned as a promising paradigm to handle the massive volume of data of COVID-19 patients using artificial intelligence (AI). Traditionally, AI techniques often require centralized data collection and training the model in a single organization, which is most common weakness due to the privacy and security of raw data communication. To solve this challenging task, we propose a blockchain-based federated learning framework that provides collaborative data training solutions by coordinating multiple hospitals to train and share encrypted federated models without leakage of data privacy. The blockchain ledger technology provides the decentralization of federated learning model without any central server. The proposed homomorphic encryption scheme encrypts and decrypts the gradients of model to preserve the privacy. More precisely, the proposed framework: i) train the local model by a novel capsule network to segmentation and classify COVID-19 images, ii) then use the homomorphic encryption scheme to secure the local model that encrypts and decrypts the gradients, and finally the model is shared over a decentralized platform through proposed blockchain-based federated learning algorithm. The integration of blockchain and federated learning leads to a new paradigm for medical image data sharing in the decentralized network. The conducted experimental resultsdemonstrate the performance of the proposed scheme.

COVID-19 Outbreak and Voluntary Demand for Non-COVID-19 Healthcare: Evidence from Taiwan (preprint)

This paper examines the impact of the COVID-19 outbreak on voluntary demand for Non-COVID-19 healthcare. We use 2014-2020 weekly county-level data from Taiwan National Health Insurance alongside a difference-in-differences design. Our results indicate that even if there are no government restrictions on human mobility, people spontaneously reduce their demand for healthcare due to fears of infection or improved health status. On average, the number of outpatient visits (inpatient admissions) decreased by 18% (9%) after COVID-19 outbreak. Furthermore, the demand response of healthcare for infectious diseases (e.g. flu) is much greater and more persistent than for other diseases, suggesting that the substantial decline in healthcare use is induced by positive public-health externality of prevention measures for COVID-19.

COVID-­19 Outbreak and Voluntary Demand for Non-­COVID­-19 Healthcare: Evidence from Taiwan (preprint)

This paper examines the impact of the COVID­-19 outbreak on voluntary demand for Non­-COVID-19 healthcare. We use 2014–2020 weekly county ­level data from Taiwan National Health Insurance alongside a difference ­in­ differences design. Our results indicate that even if there are no government restrictions on human mobility, people spontaneously reduce their demand for healthcare due to fears of infection or improved health status. On average, the number of outpatient visits (inpatient admissions) decreased by 18% (9%) after COVID­19 outbreak. Furthermore, the demand response of healthcare for infectious diseases (e.g. flu) is much greater and more persistent than for other diseases, suggesting that the substantial decline in healthcare use is induced by positive public ­health externality of prevention measures for COVID-­19.

30-day outcomes in hip fracture patients during the COVID-19 pandemic compared to the preceding year.

AIMS: To establish if COVID-19 has worsened outcomes in patients with AO 31 A or B type hip fractures. METHODS: Retrospective analysis of prospectively collected data was performed for a five-week period from 20 March 2020 and the same time period in 2019. The primary outcome was mortality at 30 days. Secondary outcomes were COVID-19 infection, perioperative pulmonary complications, time to theatre, type of anaesthesia, operation, grade of surgeon, fracture type, postoperative intensive care admission, venous thromboembolism, dislocation, infection rates, and length of stay. RESULTS: In all, 76 patients with hip fractures were identified in each group. All patients had 30-day follow-up. There was no difference in age, sex, American Society of Anesthesiologists (ASA) classification or residence at time of injury. However, three in each group were not fit for surgery. No significant difference was found in 30-day mortality; ten patients (13%) in 2019 and 11 patients (14%) in 2020 (p = 0.341). In the 2020 cohort, ten patients tested positive for COVID-19, two (20%) of whom died. There was no significant increase in postoperative pulmonary complications. Median time to theatre was 20 hours (interquartile range (IQR) 16 to 25) in 2019 versus 23 hours (IQR 18 to 30) in 2020 (p = 0.130). Regional anaesthesia increased from 24 (33%) cases in 2019 to 46 (63%) cases in 2020, but ten (14%) required conversion to general anaesthesia. In both groups, 53 (70%) operations were done by trainees. Hemiarthroplasty for 31 B type fractures was the most common operation. No significant difference was found for intensive care admission or 30-day venous thromboembolism, dislocation or infection, or length of stay. CONCLUSION: Little information exists on mortality and complications after hip fracture during the COVID-19 pandemic. At the time of writing, no other study of outcomes in the UK has been published.Cite this article: Bone Joint Open 2020;1-7:415-419.

A Sensitive, Rapid, and Portable CasRx-based Diagnostic Assay for SARS-CoV2 (preprint)

Since its first emergence from China in late 2019, the SARS-CoV-2 virus has spread globally despite unprecedented containment efforts, resulting in a catastrophic worldwide pandemic. Successful identification and isolation of infected individuals can drastically curtail virus spread and limit outbreaks. However, during the early stages of global transmission, point-of-care diagnostics were largely unavailable and continue to remain difficult to procure, greatly inhibiting public health efforts to mitigate spread. Furthermore, the most prevalent testing kits rely on reagent- and time-intensive protocols to detect viral RNA, preventing rapid and cost-effective diagnosis. Therefore the development of an extensive toolkit for point-of-care diagnostics that is expeditiously adaptable to new emerging pathogens is of critical public health importance. Recently, a number of novel CRISPR-based diagnostics have been developed to detect COVID-19. Herein, we outline the development of a CRISPR-based nucleic acid molecular diagnostic utilizing a Cas13d ribonuclease derived from Ruminococcus flavefaciens (CasRx) to detect SARS-CoV-2, an approach we term SENSR (Sensitive Enzymatic Nucleic-acid Sequence Reporter). We demonstrate SENSR robustly detects SARS-CoV-2 sequences in both synthetic and patient-derived samples by lateral flow and fluorescence, thus expanding the available point-of-care diagnostics to combat current and future pandemics.

Rhabdomyolysis is Associated with Hospital Mortality in Patients with COVID-19 (preprint)

Purpose: To evaluate the clinical features and outcomes of rhabdomyolysis (RM) in patients with COVID-19. Method: A single center retrospective cohort study of 1,014 consecutive hospitalized patients with confirmed COVID-19 at the Huoshenshan hospital in Wuhan, China, between February 17 and April 12, 2020. Results: : The overall incidence of RM was 2.2%. Comparing with patients without RM, patients with RM tended to have a higher risk of deterioration, representing by higher ratio to be admitted to the intensive care unit (ICU) (90.9 % vs 5.3%, P <0.001), and to undergo mechanical ventilation (86.4 % vs 2.7% P <0.001). Compared with patients without RM, patients with RM had laboratory test abnormalities, including indicators of inflammation, coagulation activation and kidney injury. Patients with RM had a higher risk of hospital death ( P < 0.001). Cox proportional hazard regression model confirmed that RM indicators, including peak creatine kinase (CK) >1000 IU/L (HR=6.46, 95% CI: 3.02-13.86), peak serum myoglobin (MYO) >1000 ng/mL (HR=9.85, 95% CI: 5.04-19.28) were independent risk factors for in-hospital death. Additionally, patients with COVID-19 that developed RM tended to have a delayed virus clearance. Conclusion: RM might be an important factor contributing to adverse outcomes of patients with COVID-19. Early detection and effective intervention of RM may help reduce deaths of patients with COVID-19.

Virus-free and live-cell visualizing SARS-CoV-2 cell entry for studies of neutralizing antibodies and compound inhibitors (preprint)

The ongoing COVID-19 pandemic, caused by SARS-CoV-2 infection, has resulted in hundreds of thousands of deaths. Cellular entry of SARS-CoV-2, which is mediated by the viral spike protein and host ACE2 receptor, is an essential target for the development of vaccines, therapeutic antibodies, and drugs. Using a mammalian cell expression system, we generated a recombinant fluorescent protein (Gamillus)-fused SARS-CoV-2 spike trimer (STG) to probe the viral entry process. In ACE2-expressing cells, we found that the STG probe has excellent performance in the live-cell visualization of receptor binding, cellular uptake, and intracellular trafficking of SARS-CoV-2 under virus-free conditions. The new system allows quantitative analyses of the inhibition potentials and detailed influence of COVID-19-convalescent human plasmas, neutralizing antibodies and compounds, providing a versatile tool for high-throughput screening and phenotypic characterization of SARS-CoV-2 entry inhibitors. This approach may also be adapted to develop a viral entry visualization system for other viruses.

Blockchain-Federated-Learning and Deep Learning Models for COVID-19 detection using CT Imaging (preprint)

With the increase of COVID-19 cases worldwide, an effective way is required to diagnose COVID-19 patients. The primary problem in diagnosing COVID-19 patients is the shortage and reliability of testing kits, due to the quick spread of the virus, medical practitioners are facing difficulty identifying the positive cases. The second real-world problem is to share the data among the hospitals globally while keeping in view the privacy concerns of the organizations. Building a collaborative model and preserving privacy are major concerns for training a global deep learning model. This paper proposes a framework that collects a small amount of data from different sources (various hospitals) and trains a global deep learning model using blockchain based federated learning. Blockchain technology authenticates the data and federated learning trains the model globally while preserving the privacy of the organization. First, we propose a data normalization technique that deals with the heterogeneity of data as the data is gathered from different hospitals having different kinds of CT scanners. Secondly, we use Capsule Network-based segmentation and classification to detect COVID-19 patients. Thirdly, we design a method that can collaboratively train a global model using blockchain technology with federated learning while preserving privacy. Additionally, we collected real-life COVID-19 patients data, which is, open to the research community. The proposed framework can utilize up-to-date data which improves the recognition of computed tomography (CT) images. Finally, our results demonstrate a better performance to detect COVID-19 patients.