Characterization of spike processing and entry mechanisms of seasonal human coronaviruses NL63, 229E and HKU1 (preprint)

Although much has been learned about the entry mechanism of severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), the details of entry mechanisms of seasonal human coronaviruses (HCoVs) remain less well understood. In the present study, we established that 293T cell lines that stably express angiotensin converting enzyme (ACE2), aminopeptidase N (APN), or transmembrane serine protease 2 (TMPRSS2) support high level transduction of lentiviral pseudoviruses bearing spike proteins of seasonal HCoVs, HCoV-NL63, -229E, or -HKU1, respectively. Our results showed that entry of HCoV-NL63, -229E and -HKU1 pseudoviruses is sensitive to endosomal acidification inhibitors (chloroquine and NH4Cl), indicating virus entry via the endocytosis route. Although HCoV-HKU1 pseudovirus infection requires TMPRSS2 expression on cell surface, endocytosis-mediated HCoV-HKU1 entry requires the serine protease domain but not the serine protease activity of TMPRSS2. We also show that amino acids in the predicted S1/S2 junctions of spike proteins of HCoV-NL63, and -229E are essential for optimal entry but non-essential for spike-mediated entry of HCoV-HKU1. Our findings provide insights into entry mechanism of seasonal HCoVs that may support the development of novel treatment strategies.

Antigenic cartography using hamster sera identifies SARS-CoV-2 JN.1 evasion seen in human XBB.1.5 booster sera (preprint)

Antigenic assessments of SARS-CoV-2 variants inform decisions to update COVID-19 vaccines. Primary infection sera are often used for assessments, but such sera are rare due to population immunity from SARS-CoV-2 infections and COVID-19 vaccinations. Here, we show that neutralization titers and breadth of matched human and hamster pre-Omicron variant primary infection sera correlate well and generate similar antigenic maps. The hamster antigenic map shows modest antigenic drift among XBB sub-lineage variants, with JN.1 and BA.4/BA.5 variants within the XBB cluster, but with five to six-fold antigenic differences between these variants and XBB.1.5. Compared to sera following only ancestral or bivalent COVID-19 vaccinations, or with post-vaccination infections, XBB.1.5 booster sera had the broadest neutralization against XBB sub-lineage variants, although a five-fold titer difference was still observed between JN.1 and XBB.1.5 variants. These findings suggest that antibody coverage of antigenically divergent JN.1 could be improved with a matched vaccine antigen.

Event Detection from Social Media for Epidemic Prediction (preprint)

Social media is an easy-to-access platform providing timely updates about societal trends and events. Discussions regarding epidemic-related events such as infections, symptoms, and social interactions can be crucial for informing policymaking during epidemic outbreaks. In our work, we pioneer exploiting Event Detection (ED) for better preparedness and early warnings of any upcoming epidemic by developing a framework to extract and analyze epidemic-related events from social media posts. To this end, we curate an epidemic event ontology comprising seven disease-agnostic event types and construct a Twitter dataset SPEED with human-annotated events focused on the COVID-19 pandemic. Experimentation reveals how ED models trained on COVID-based SPEED can effectively detect epidemic events for three unseen epidemics of Monkeypox, Zika, and Dengue; while models trained on existing ED datasets fail miserably. Furthermore, we show that reporting sharp increases in the extracted events by our framework can provide warnings 4-9 weeks earlier than the WHO epidemic declaration for Monkeypox. This utility of our framework lays the foundations for better preparedness against emerging epidemics.

Causal Graph ODE: Continuous Treatment Effect Modeling in Multi-agent Dynamical Systems (preprint)

Real-world multi-agent systems are often dynamic and continuous, where the agents co-evolve and undergo changes in their trajectories and interactions over time. For example, the COVID-19 transmission in the U.S. can be viewed as a multi-agent system, where states act as agents and daily population movements between them are interactions. Estimating the counterfactual outcomes in such systems enables accurate future predictions and effective decision-making, such as formulating COVID-19 policies. However, existing methods fail to model the continuous dynamic effects of treatments on the outcome, especially when multiple treatments (e.g., "stay-at-home" and "get-vaccine" policies) are applied simultaneously. To tackle this challenge, we propose Causal Graph Ordinary Differential Equations (CAG-ODE), a novel model that captures the continuous interaction among agents using a Graph Neural Network (GNN) as the ODE function. The key innovation of our model is to learn time-dependent representations of treatments and incorporate them into the ODE function, enabling precise predictions of potential outcomes. To mitigate confounding bias, we further propose two domain adversarial learning-based objectives, which enable our model to learn balanced continuous representations that are not affected by treatments or interference. Experiments on two datasets (i.e., COVID-19 and tumor growth) demonstrate the superior performance of our proposed model.

CD4+ and CD8+ T cells are required to prevent SARS-CoV-2 persistence in the nasal compartment (preprint)

SARS-CoV-2 is the causative agent of COVID-19 and continues to pose a significant public health threat throughout the world. Following SARS-CoV-2 infection, virus-specific CD4+ and CD8+ T cells are rapidly generated to form effector and memory cells and persist in the blood for several months. However, the contribution of T cells in controlling SARS-CoV-2 infection within the respiratory tract are not well understood. Using C57BL/6 mice infected with a naturally occurring SARS-CoV-2 variant (B.1.351), we evaluated the role of T cells in the upper and lower respiratory tract. Following infection, SARS-CoV-2-specific CD4+ and CD8+ T cells are recruited to the respiratory tract and a vast proportion secrete the cytotoxic molecule Granzyme B. Using antibodies to deplete T cells prior to infection, we found that CD4+ and CD8+ T cells play distinct roles in the upper and lower respiratory tract. In the lungs, T cells play a minimal role in viral control with viral clearance occurring in the absence of both CD4+ and CD8+ T cells through 28 days post-infection. In the nasal compartment, depletion of both CD4+ and CD8+ T cells, but not individually, results in persistent and culturable virus replicating in the nasal compartment through 28 days post-infection. Using in situ hybridization, we found that SARS-CoV-2 infection persisted in the nasal epithelial layer of tandem CD4+ and CD8+ T cell-depleted mice. Sequence analysis of virus isolates from persistently infected mice revealed mutations spanning across the genome, including a deletion in ORF6. Overall, our findings highlight the importance of T cells in controlling virus replication within the respiratory tract during SARS-CoV-2 infection.

A moderately higher time-in-range threshold improves the prognosis of type-2 diabetes patients complicated with COVID-19 (preprint)

After fully lifting coronavirus disease 2019 (COVID-19) pandemic control measures in mainland China in 12/2022, the incidence of COVID-19 has increased markedly, making it difficult to meet the general time-in-range (TIR) requirement. We investigated a more clinically practical TIR threshold and examined its association with the prognosis of COVID-19 patients with type-2 diabetes. Sixty-three type-2 diabetes patients complicated with COVID-19 were evaluated. Patient information included epidemiological and laboratory characteristics, treatment options and outcomes. The percentages of time-above-range (TAR), time-below-range (TBR) and TIR were calculated from intermittently scanned continuous glucose monitoring. The composite end point included a >20-day length of stay, intensive care unit admission, mechanical ventilation use, or death. TIR with thresholds of 80 to 190 mg/dL was significantly associated with favorable outcomes. An increase of 1% in TIR is connected with a reduction of 3.70% in the risk of adverse outcomes. The Youden index was highest when the TIR was 54.73%, and the sensitivity and specificity were 58.30% and 77.80%, respectively. After accounting for confounding variables, our analysis revealed that threshold target ranges (TARs) ranging from 200 mg/dL to 230 mg/dL significantly augmented the likelihood of adverse outcomes.The TIR threshold of 80 to 190 mg/dL has a comparatively high predictive value of the prognosis of COVID-19. TIR >54.73% was associated with a decreased risk of adverse outcomes. These findings provide clinically critical insights into possible avenues to improve outcomes for COVID-19 patients with type-2 diabetes.

Decoding Susceptibility: Modeling Misbelief to Misinformation Through a Computational Approach (preprint)

Susceptibility to misinformation describes the degree of belief in unverifiable claims, a latent aspect of individuals' mental processes that is not observable. Existing susceptibility studies heavily rely on self-reported beliefs, which can be subject to bias, expensive to collect, and challenging to scale for downstream applications. To address these limitations, in this work, we propose a computational approach to model users' latent susceptibility levels. As shown in previous research, susceptibility is influenced by various factors (e.g., demographic factors, political ideology), and directly influences people's reposting behavior on social media. To represent the underlying mental process, our susceptibility modeling incorporates these factors as inputs, guided by the supervision of people's sharing behavior. Using COVID-19 as a testbed domain, our experiments demonstrate a significant alignment between the susceptibility scores estimated by our computational modeling and human judgments, confirming the effectiveness of this latent modeling approach. Furthermore, we apply our model to annotate susceptibility scores on a large-scale dataset and analyze the relationships between susceptibility with various factors. Our analysis reveals that political leanings and psychological factors exhibit varying degrees of association with susceptibility to COVID-19 misinformation.

MIDDAG: Where Does Our News Go? Investigating Information Diffusion via Community-Level Information Pathways (preprint)

We present MIDDAG, an intuitive, interactive system that visualizes the information propagation paths on social media triggered by COVID-19-related news articles accompanied by comprehensive insights, including user/community susceptibility level, as well as events and popular opinions raised by the crowd while propagating the information. Besides discovering information flow patterns among users, we construct communities among users and develop the propagation forecasting capability, enabling tracing and understanding of how information is disseminated at a higher level.

Comparative Analysis of SARS-CoV-2 Antigenicity across Assays and in Human and Animal Model Sera (preprint)

The antigenic evolution of SARS-CoV-2 requires ongoing monitoring to judge the immune escape of newly arising variants. A surveillance system necessitates an understanding of differences in neutralization titers measured in different assays and using human and animal sera. We compared 18 datasets generated using human, hamster, and mouse sera, and six different neutralization assays. Titer magnitude was lowest in human, intermediate in hamster, and highest in mouse sera. Fold change, immunodominance patterns and antigenic maps were similar among sera. Most assays yielded similar results, except for differences in fold change in cytopathic effect assays. Not enough data was available for conclusively judging mouse sera, but hamster sera were a consistent surrogate for human first-infection sera.

Immune and behavioral correlates of protection against symptomatic post-vaccination SARS-CoV-2 infection (preprint)

Background: We sought to determine immune and behavioral pre-infection correlates of protection against SARS-CoV-2 post-vaccine infections in a joint analysis of epidemiological and immunological cohort data. Methods: Serum and saliva samples from 176 BNT162b2-vaccinated adults in the Prospective Assessment of SARS-CoV-2 Seroconversion study were collected between October and December 2021 and assessed for serum and saliva levels of Wuhan-1 wild-type (WT) SARS-CoV-2 Spike (S)-specific IgG and IgA binding antibodies (bAb) using a multiplex microsphere-based immunoassay (MMIA). Serum samples were also assessed for WT receptor binding domain (RBD)-specific bAb by two commercial assays, BA.1 S-specific IgG bAb by MMIA, and neutralization activity against D614G, Delta (B.1.617.2), and Omicron BA.1 and BA.1.1 variants using a lentiviral pseudovirus neutralization assay. After the Fall 2021 visit, participants reported all positive PCR and/or antigen tests for SARS-CoV-2. Duration, severity, and type of symptoms, as well as risk exposures and adherence to precautionary measures, were assessed by questionnaires during the Spring 2022 visit. Results: Thirty-two participants (18.2%) developed symptomatic post-vaccination SARS-CoV-2 infections (PVI) between December 7, 2021 and April 1, 2022. Pre-infection WT (geometric mean (GM) of 3,863 vs 2,736 binding antibody unit [BAU]/ml, uninfected vs PVI, p=0.0098) and BA.1 (GM of 276.9 vs 179.9 arbitrary bAb unit [AU]/ml, uninfected vs PVI, p=0.04) anti-S IgG bAb levels measured by MMIA and neutralizing titers (NT) against BA.1 (GM titer [GMT] of 493.6 vs 286.2, uninfected vs PVI, p=0.0313) and BA.1.1 (GMT of 552.0 vs 302.5, uninfected vs PVI, p=0.021) were significantly higher in individuals that did not develop PVIs. WT anti-S bAb levels greater than 5,000 BAU/ml were associated with > 90% protection against symptomatic PVI. In individuals that developed PVI, WT anti-S IgG bAb levels correlated with lower disease severity scores ({rho}= -0.3859, p=0.032) and shorter duration of clinical disease ({rho}= -0.5273, p=0.0023). WT anti-RBD bAb levels measured by commercial assays correlated strongly with bAb levels measured by MMIA ({rho}=0.8239, p<0.0001 and {rho}=0.6929, p<0.0001, Roche and Siemens assays, respectively), but did not reach statistical significance for correlation with protection against PVI. Home risk score, but neither work nor home precautionary measures, correlated strongly with risk of PVI (mean score of 20.77 vs 47.33, uninfected vs PVI respectively, p<0.0001). Conclusions: Anti-S IgG bAb levels (directed against either WT or Omicron BA.1 subvariant) and NTs served as correlates of protection against symptomatic SARS-CoV-2 infection. Anti-S (WT) IgG bAb levels remained a significant correlate of protection against PVIs when adjusting for demography and risk behavior. Results of this study also suggest that commercial assays for anti-S bAb may need to be reformatted to enable detection of higher maximum values for use as predictors of increased susceptibility to SARS-CoV-2 infection.

D614G and Omicron SARS-CoV-2 variant spike proteins differ in the effects of N-glycan modifications on spike expression, virus infectivity, and neutralization by some therapeutic antibodies (preprint)

The SARS-CoV-2 spike glycoprotein has 22 potential N-linked glycosylation sites per monomer that are highly conserved among diverse variants, but how individual glycans affect virus entry and neutralization of Omicron variants has not been extensively characterized. Here we compared the effects of specific glycan deletions or modifications in the Omicron BA.1 and D614G spikes on spike expression, processing, and incorporation into pseudoviruses, as well as on virus infectivity and neutralization by therapeutic antibodies. We found that loss of potential glycans at spike residues N717 and N801 each conferred a loss of pseudovirus infectivity for Omicron but not for D614G or Delta variants. This decrease in infectivity correlated with decreased spike processing and incorporation into Omicron pseudoviruses. Oligomannose-enriched Omicron pseudoviruses generated in GnTI- cells or in the presence of kifunensine were non-infectious, whereas D614G or Delta pseudoviruses generated under similar conditions remained infectious. Similarly, authentic SARS-CoV-2 grown in the presence of kifunensine decreased titers more for the BA.1.1 variant than Delta or D614G variants relative to their respective, untreated controls. Finally, we found that loss of some N-glycans, including N343 and N234, increased the maximum percent neutralization by the class 3 S309 monoclonal antibody against D614G but not BA.1 variants, while these glycan deletions altered the neutralization potency of the class 1 COV2-2196 and Etesevimab monoclonal antibodies without affecting maximum percent neutralization. The maximum neutralization by some antibodies also varied with the glycan composition, with oligomannose-enriched pseudoviruses conferring the highest percent neutralization. These results highlight differences in the interactions between spike glycans and residues among SARS-CoV-2 variants that can affect spike expression, virus infectivity, and susceptibility of variants to antibody neutralization.

Nowcasting bitcoin's crash risk with order imbalance

The spectacular nature of bitcoin price crashes baffles market spectators and prompts routine warnings from regulators cautioning that cryptocurrencies behave in contra to the fundamental properties that traditionally define what constitutes money. Arguably most concerning to the public is, first, bitcoin's unprecedented price volatility relative to other asset classes and, second, its seemingly detached price behavior relative to time-honored economic and market fundamentals. In an attempt to create an early warning system of bitcoin price crash risk using generalized extreme value (GEV) and logistic regression modeling, this study integrates order flow imbalance, along with several control factors which reflect blockchain activity and network value, in order to nowcast bitcoin's price crashes. From a data analysis perspective, and despite their dissimilar distributional underpinnings, the GEV and logistic models perform comparably. When evaluating the type I and type II errors which these models yield, it is shown that their performance is comparable in terms of accuracy. In addition, it is also shown how the proportion of type I and type II errors can shift dramatically across probability cutoff tolerances. Towards the end of this study, time varying probabilities of a price crash are shown and evaluated. The sample range in this study encompasses the SARS-CoV-2 (Covid-19) time period as well as the recent scandal and collapse of the FTX cryptocurrency exchange.

A Decade of Scientific and Technological Innovation at Chinese Academy of Medical Sciences & Peking Union Medical College: Retrospect and Prospect.

Focusing on the reform initiatives of Chinese Academy of Medical Sciences (CAMS) & Peking Union Medical College (PUMC) in medical scientific and technological innovation from perspectives of deepening the reform and optimizing the ecosystem of science and technology innovation, this article summarizes the highlights of CAMS & PUMC's efforts in safeguarding people's health and promoting the Healthy China 2030 strategy through scientific and technological innovation in the fields including basic research, disease prevention and treatment, and medical technology in the past ten years. These achievements embody the endeavors and responsibility of CAMS & PUMC in realizing self-reliance and self-improvement of Chinese medical science and technology and highlight its contributions to the development of medical science and technology of China.

Bivalent COVID-19 vaccine antibody responses to Omicron variants suggest that responses to divergent variants would be improved with matched vaccine antigens (preprint)

We compared neutralizing antibody responses to BA.4/5, BQ.1.1, XBB, and XBB.1.5 Omicron SARS-CoV-2 variants after a bivalent or ancestral COVID-19 mRNA booster vaccine or post-vaccination infection. We found that the bivalent booster elicited moderately high antibody titers against BA.4/5 that were approximately two-fold higher against all Omicron variants than titers elicited by the monovalent booster. The bivalent booster elicited low but similar titers against both XBB and XBB.1.5 variants. These findings inform risk assessments for future COVID-19 vaccine recommendations and suggest that updated COVID-19 vaccines containing matched vaccine antigens to circulating divergent variants may be needed.

Online flipped learning methods for teaching hospitality skills and management practices in an epidemic situation: A study on learning attitude and effectiveness

Starting in 2019, the ongoing COVID-19 pandemic has lasted 3 years and will likely continue to affect the lives of people all over the world. According to a United Nations Educational, Scientific and Cultural Organization (UNESCO) survey, more than 91% of students from all over the world have been affected by the spread of COVID-19. The application of technological networks can help solve problems related to being unable to attend school in person, as online teaching can effectively help reduce learning loss in the short term. In Taiwan, the higher education system has been using online learning, but now faces a new and huge crisis, as some courses do not readily translate to this setting. In professional courses run by hospitality departments, it is essentially impossible to accurately convey the practical skills required, for example, aspects of color, aroma, and taste through online teaching. Moreover, the learning level of each student varies greatly. During the online teaching process, instructors teach professional skills and movements through a single teaching video, which may not meet the needs of all students. In response, this study explores using the flipped teaching method, to not only enable students to master and control their learning and effectively adjust their self- adaptive learning progress but also to help teachers solve problems and impart professional skills using a two-way, interactive, online teaching method. This approach, flipping a class in an online learning environment, could effectively make up for the one-way teaching sometimes created by video content, and address the problem of gaps in learning professional practical skills. It can also induce students with poor learning attitudes to actively participate in learning. This study involved 55 bachelor students from a university of science and technology in Taiwan. The research results are as follows: (1) Students who participated in the flipped teaching mode, which involved two-way interaction showed better professional understanding of the course and improved willingness to learn, thereby improving the learning effect. (2) Awareness of these poor practical catering professional skills in students, assisted in laying the professional foundation for students to gradually improve their learning attitude and their advanced skills. This indicates that students with poor academic performance in an online environment might benefit from two-way interactive teaching. Teachers should clarify detailed descriptions of professional practical actions that confuse students. (3) In flipped learning, the grouping of "game/toy-based e-learning” can not only improve the performance of students who actively study to achieve good grades but also help and motivate other students to learn together. These results indicate that in flipped classrooms that use an online learning environment, the active learning and learning attitudes of students were positive and that their interest in learning and learning efficiency was also significantly improved. At the same time, this approach stimulated the innovation, creativity, and creative development of students in using professional technology in the hospitality industry. It transformed the passive learning situation of online one-way teaching into an active two-way teaching environment.

Bivalent mRNA vaccine improves antibody-mediated neutralization of many SARS-CoV-2 Omicron lineage variants (preprint)

The early Omicron lineage variants evolved and gave rise to diverging lineages that fueled the COVID-19 pandemic in 2022. Bivalent mRNA vaccines, designed to broaden protection against circulating and future variants, were authorized by the U.S. Food and Drug Administration (FDA) in August 2022 and recommended by the U.S. Centers for Disease Control and Prevention (CDC) in September 2022. The impact of bivalent vaccination on eliciting neutralizing antibodies against homologous BA.4/BA.5 viruses as well as emerging heterologous viruses needs to be analyzed. In this study, we analyze the neutralizing activity of sera collected after a third dose of vaccination (2-6 weeks post monovalent booster) or a fourth dose of vaccination (2-7 weeks post bivalent booster) against 10 predominant/recent Omicron lineage viruses including BA.1, BA.2, BA.5, BA.2.75, BA.2.75.2, BN.1, BQ.1, BQ.1.1, XBB, and XBB.1. The bivalent booster vaccination enhanced neutralizing antibody titers against all Omicron lineage viruses tested, including a 10-fold increase in neutralization of BQ.1 and BQ.1.1 viruses that predominated in the U.S. during the last two months of 2022. Overall, the data indicate the bivalent vaccine booster strengthens protection against Omicron lineage variants that evolved from BA.5 and BA.2 progenitors.

Maternal and infant outcomes in women with and without gestational diabetes mellitus in the COVID-19 era in China: Lessons learned

Aims The global COVID-19 pandemic has required a drastic transformation of prenatal care services. Whether the reformulation of the antenatal care systems affects maternal and infant outcomes remains unknown. Particularly, women with gestational diabetes mellitus (GDM) are among those who bear the greatest brunt. Thus, this study aimed to evaluate the impact of COVID-19 lockdown during late pregnancy on maternal and infant outcomes in women stratified by the GDM status in China. Study design The participants were women who experienced the COVID-19 lockdown during late pregnancy (3185 in the 2020 cohort) or not (2540 in the 2019 cohort) that were derived from the Beijing Birth Cohort Study. Maternal metabolic indicators, neonatal outcomes, and infant anthropometrics at 12 months of age were compared between the two cohorts, stratified by the GDM status. Results Participants who experienced COVID-19 lockdown in late pregnancy showed lower gestational weight gain than those in the control cohort. Nevertheless, they displayed a worse metabolic profile. COVID-19 lockdown during pregnancy was associated with higher glycosylated hemoglobin (HbA1c) (β= 0.11, 95% CI = 0.05–0.16, q-value = 0.002) and lower high density lipoprotein cholesterol level (HDL-C) level (β=–0.09, 95% CI = –0.14 to –0.04, q-value = 0.004) in women with GDM, adjusted for potential confounders. In normoglycemic women, COVID-19 lockdown in late pregnancy was associated with higher fasting glucose level (β= 0.10, 95% CI = 0.08–0.12, q-value <0.0001), lower HDL-C level (β=–0.07, 95% CI = –0.08 to –0.04, q-value <0.0001), and increased risk of pregnancy-induced hypertension (adjusted OR=1.80, 95%CI=1.30–2.50, q-value=0.001). The fasting glucose level decreased less from early to late pregnancy in women who experienced COVID-19 lockdown than in the controls, regardless of the GDM status. The HDL-C has risen less with COVID-19 lockdown in the normoglycemic subgroup. In contrast, no significant differences regarding neonatal outcomes or infant weight were found between the two cohorts. Conclusion Experiencing the COVID-19 lockdown in pregnancy was associated with worse maternal metabolic status but similar neonatal outcomes and infant weight.

The machine learning model based on trajectory analysis of ribonucleic acid test results predicts the necessity of quarantine in recurrently positive patients with SARS-CoV-2 infection

Background SARS-CoV-2 patients re-experiencing positive nucleic acid test results after recovery is a concerning phenomenon. Current pandemic prevention strategy demands the quarantine of all recurrently positive patients. This study provided evidence on whether quarantine is required in those patients, and predictive algorithms to detect subjects with infectious possibility. Methods This observational study recruited recurrently positive patients who were admitted to our shelter hospital between May 12 and June 10, 2022. The demographic and epidemiologic data was collected, and nucleic acid tests were performed daily. virus isolation was done in randomly selected cases. The group-based trajectory model was developed based on the cycle threshold (Ct) value variations. Machine learning models were validated for prediction accuracy. Results Among the 494 subjects, 72.04% were asymptomatic, and 23.08% had a Ct value under 30 at recurrence. Two trajectories were identified with either rapid (92.24%) or delayed (7.76%) recovery of Ct values. The latter had significantly higher incidence of comorbidities;lower Ct value at recurrence;more persistent cough;and more frequently reported close contacts infection compared with those recovered rapidly. However, negative virus isolation was reported in all selected samples. Our predictive model can efficiently discriminate those with delayed Ct value recovery and infectious potentials. Conclusion Quarantine seems to be unnecessary for the majority of re-positive patients who may have low transmission risks. Our predictive algorithm can screen out the suspiciously infectious individuals for quarantine. These findings may assist the enaction of SARS-CoV-2 pandemic prevention strategies regarding recurrently positive patients in the future.