COVID-19 Virtual Symposium #16

The sixteenth Columbia COVID-19 virtual symposium

By
VP&S Office for Research
July 29, 2020

Today was the sixteenth Columbia COVID-19 Virtual Symposium.

Presentation Summaries and Video Recordings

Written summaries were kindly provided by the Presenters or the Columbia Researchers Against COVID-19 (CRAC) Team.

If you are having trouble viewing the embedded videos, try refreshing the page. If that doesn't work, visit the Herbert Irving Comprehensive Cancer Center's YouTube page

Talk 1: Juan Francisco Saldarriaga Chaux, MS, MArch - Brown Institute for Media Innovation, Columbia University

Title: Moving Around During the Pandemic — Analyzing Choice of Transportation in New York City During COVID-19

Summary: Presented by the Brown Institute of Media Innovation, this talk presented a study on the transportation system in New York City. Using an existing dataset, this study analyzed change in ridership as compared to median household income. Household income was positively related to the decreased ridership in a neighborhood. Compared to 2019, the findings showed a dramatic decrease in subway use during the COVID-19 period. Citi Bike use during the pandemic initially decreased, then steadily rose, compared to the same period in 2019. The bike count during the pandemic dropped, then recovered compared to the similar time period in the year 2019. The research has ongoing data analysis to understand disaggregated Citi Bike trips by time of day, trip duration, and the spatial distribution of the Citi Bike usage. Further analysis will be conducted to analyze Citi Bike usage by the length of trip, subway usage by time of day, and bike counters by the time of day. Other relevant data will be incorporated into the overall analysis, including bus, scooters, demographics (income and race), land use, economic activity, and other mobility data.

Juan Francisco Saldarriaga Chaux's presentation

Talk 2: Kartik Chandran, PhD - Professor, Department of Microbiology & Immunology, Albert Einstein College of Medicine

Title: Detecting, Dissecting, and Disabling the SARS-CoV-2 spike

Summary: To develop a serologic assay for SARS-CoV-2 spike specific antibodies, there was an acute need for COVID-19 antibody testing for patients and HCWs. However, barriers exist, including the shifting landscape of commercial tests, sensitivity/specificity not always clear, hard to “future-proof” given the high demand, and limited supply. A group of research collaborators has produced SARS-CoV-2 spike protein at scale and turned ELISA curves into a single cut-off value for Dx. The preliminary results showed assay was highly reproducible. After selected cut-offs to maximize specificity and sensitivity, a spike-based Dx test distinguishes SARS2 from hCoVs. The current research has validated against the Wadworth NYS test, validated against Columbia LDT, conducted a large cross-validation study at MMC almost complete, and applied for NYS emergency use authorization. Also, the assay has been ported to automated format in the MMC Pathology CLIA-certified lab and cross-validated. Therefore, to analyze the BSL-2 assays, there was ongoing demand to study viral infection mechanisms, screen small-molecule inhibitors for antiviral activity, profile the immunological activity of candidate vaccines, and screen donor plasma and monoclonal antibodies for neutralizing activity. It turned out that convalescent plasma appeared to have value in improving outcomes in moderate-severely ill COVID-19 patients. Antibody therapeutics have shown to be effective against other viral diseases. There is ongoing research to analyze whether the same approach can be applied to COVID-19.

Kartik Chandran's presentation

Talk 3: Ponisseril Somasundaran, PhD - LVD Krumb Professor, Earth and Environmental Engineering, Columbia University

Title: Green Disinfectant Foams to mitigate/kill Virus/Microbes Spreading

Summary: To start the lecture, Dr. Somasundaran introduced that disinfectant products used for COVID-19 may increase the risk of chronic lung diseases. However, challenges exist in current decontamination practices, for instance, zip bags with chlorine dioxide for corpse decontamination, toxic wash seeps to ground, and it is hard to get crevices. In dry foam, the volume occupied by the liquid material is small. Also, the foam can penetrate into cracks with functionalized silicone super spreaders. Currently, the formulation is being prepared with synthetic and greener surfactants to control foam texture. Foam stability is controlled by repulsion between Lamallaes and drainage. In the past, the uniform deposition of foam on a surface was assessed by using a chlorine indicator; in the future, it will be by conducting the spectroscopic evaluation of the surface treated with foam formulation. To replicate laboratory observation for foam properties with commercially available foam sprayers/generators, the current study aimed to assess foam texture on a surface from the time of deposition to its complex breakage. The guidelines included the use of chlorine indicator to assess uniformity in disinfectant deposition on the surface and spectroscopic assessment of uniformity in disinfectant deposition on the surface. The foam spray benefits including no mist, decontamination of difficult-to-reach areas, better adherence for the desired time, minimal splashback, and less water and less seepage. Therefore, consumers and regulators need surfactants to be made from natural sources with green technology. Green disinfection foams have shown the potential in multi-faceted ways and in killing existing viruses.

Ponisseril Somasundaran's presentation

Talk 4: Vishal Misra, PhD - Professor, Dept. of Computer Science, Columbia University

Title: A synthetic controls analysis of post Memorial Day COVID outbreak in the US

Summary: Synthetic controls is an empirical methodology for causal inference using observational data. In this analysis, a synthetic version of the treatment unit is created via untreated donor units linearly and the correlation between different units, i.e. synthetic and actual data is exploited. Virus reached different areas at different times, and for the purposes of COVID-19 analysis, time series aren’t aligned in terms of absolute dates. Therefore, the data can be aligned based on lockdown dates provided by IHME (Institute for Health Metrics and Evaluation). Using this approach, New York cases and deaths can be predicted utilizing Western Europe as a donor pool for creating synthetic versions; after 5 days of training, 120 days of predictions can be created accurately. In addition, it can analyze counterfactual analysis to predict how it would affect the number of cases and deaths if lockdown measures were implemented earlier. The spread of COVID-19 cases in the United States has 4 different patterns depending on the geography; they created cluster regions according to how mobility changed after lockdown and regions were clustered according to post-lockdown mobility data. To perform post-Memorial Day analysis, the model was trained for a county based on a donor pool of counties with intervention date of Memorial Day, the donor pool was filtered to only include counties at similar stages of COVID-19 spread and synthetic model of county with actual behavior was compared. The reasons for a deviation from predictions, for instance, predictions in Florida or Georgia, might be due to AC usage, implementing mask usage or herd immunity, i.e. when enough people are infected, the spread is slower. Overall, synthetic control analysis is a powerful technique to do counterfactual predictions enabling quick analysis of policy choices and number of cases/deaths. It can have a wide range of applications from analyzing the impact of tariffs on consumer goods, optimizing the drug discovery/clinical trials process, predicting sports scores to predicting network traffic.

Vishal Misra's presentation

Talk 5: Kenrick Cato, PhD, RN - Assistant Professor, School of Nursing, Columbia University

Title: COVID-19 – a predictive model providing a risk-based approach to improve clinical decision-making in the Emergency Department

Summary: As there were an increased number of patients during the COVID surge, the main question was who should be discharged home or admitted to inpatient. Therefore, the goal was to build a clinical decision support tool for patients suspected of COVID-19 infection. In the healthcare process models, they identify features of clinical behavior patterns in predictive models for association with outcomes. It is important to incorporate clinical domain expertise and clinician feedback for accurate interpretations. They used electronic health record data analyzing the patterns of interaction and integrate viral sign entry frequencies, orders (consults), demographics and nursing/clinician notes. The outcome of the model is binary: 1) admitted to inpatient 2) discharged from ED and didn’t return with inpatient admit within 72 hours. The statistical approach was implementing logistic regression, random forest classifier and boosted decision trees (XGBoost). Based on the preliminary analysis, the gradient boosting model (GBM) performed by far the best. The future studies will include continuing modeling, validating models with clinicians, developing clinical decision support intervention and validating interventions.

Talk 6: Ziyang Zhang, PhD - Damon Runyon Cancer Research Foundation Postdoctoral Fellow, Howard Hughes Medical Institute, University of California, San Francisco

Title: SARS-CoV-2 nsp14 Suppresses Viral Antigen Presentation by Down-Regulating Host MHC-I

Summary: Dr. Zhang gave a first glimpse into the ongoing research on how SARS-CoV-2 suppresses the antiviral MHC Class I antiviral immune system answer. The adaptive immune response of the human body is still mostly unknown so far, but the long incubation period, the prolonged disease and the fact that some patients show only a low concentration of antibody titers suggests that SARS-CoV-2 downregulates some parts of the immune system. Indeed it was found that some of the MHC Class I proteins responsible for killing infected cells are downregulated by SARS-CoV-2. One of the viral proteins found to be responsible for this reaction is the nsp14 protein for which such behaviour was unknown before. Additionally, first indications for IMPDH2 inhibitors counteracting the downregulation again have been found.

Talk 7: Casey Ta, PhD - Associate Research Scientist, Dept. of Biomedical Informatics, Columbia University

Title: COVID-19 Clinical Phenotyping Update

Summary: The Columbia Open Health Data system provides researchers with clinical statistics that have been mined from CUIMC clinical data. Information recorded includes the prevalence of conditions, drugs, and procedures. The web interface can be found at http://www.cohd.io. Data sets were collected from 3 cohorts: Hospitalized COVID-19 patients, influenza patients, and general patients. Prevalence and co-occurrence rates are calculated per visit. The COVID-19 cohort is split 50/50 between those in the 18-64 age group and 65+. Dyspnea, cough, and abdominal pain are reported in many more COVID-19 patients than over influenza patients. Finally, the Doc2Vec patient vector was used to create COVID-19 symptom maps using k-means clustering to define specific subtypes.

Casey Ta's presentation

Talk 8: Ian Wilson, DPhil - Hansen Professor of Structural Biology, Chairman, Dept. of Integrative Structural and Computational Biology California Campus, Scripps Research Institute

Title: Structural basis of SARS-CoV-2 recognition by neutralizing antibodies from convalescent patients

Summary: Spike (S) is the major protein of the virus surface and has been a common target for inducing neutralizing antibodies against SARS-CoV-2. Cryo-EM tomography of SARS-CoV-2 virions show that actually there are fewer spikes than previously depicted. Both the prefusion and post-fusion forms have different RBDs. When testing six different generated antibodies from SARS patients, only CR3022 cross-reacted with RBDs of both SARS and SARS-CoV-2, and can neutralize both. Notably, the CR3022 epitope is much more conserved than the ACE2 binding site, though the RBD epitope is only exposed in the “up” conformation.

Ian Wilson's presentation