Winter 2024 CIC Webinar Recap
Guest Post: Anna Eggers
The twenty-eighth of the COVID Information Commons (CIC) webinar series took place on January 30, 2024. In this forum, leading COVID-19 scientists presented their current research on the global pandemic.
Event moderators included Florence Hudson, Executive Director of the Northeast Big Data Innovation Hub at Columbia University and COVID Information Commons Principal Investigator (PI), Lauren Close, Operations & Communications Manager, and Emily Rothenberg, National Student Data Corps (NSDC) Program Manager.
The researchers presented on a range of topics, including mitigation strategies, neural slowing, predicting COVID’s future spread, and more. Each touched on broader themes related to the COVID-19 pandemic.
This session started with a presentation from Courtney Baird, a Ph.D. student at Brown University. Baird discussed her project, The Impact of Non-Pharmaceutical Interventions on the Rate of Growth of COVID-19. This project was funded by NIH’s National Institute on Aging.
Baird and her team were motivated by the insufficient data available about non-pharmaceutical interventions (NPIs) for the pandemic. Prior studies were limited in scope and provided mixed results regarding the effectiveness of NPIs in the U.S. at the height of COVID-19. Baird’s project widened the scope of this research by using county-level data to evaluate five different NPIs during the first four waves of the pandemic. They evaluated the impact of different policies on the spread of COVID-19, including large gathering bans, stay-at-home orders, face mask mandates, and bar and restaurant closures. They found that counties with a high volume of COVID-19 policy implementations tended to have lower contagion rates. At the beginning of the first wave, confusion and mask scarcity limited policy implementations until after several outbreaks had already occurred. By wave four (March through June 2021), many counties with high vaccination rates began to drop their NPIs, due to both reduced risk of serious infection as well as lower compliance rates due to pandemic fatigue. Overall, Baird and her team found that the degree of NPI effectiveness depended on timing, dosage, and incentives to compliance.
A video of Baird's presentation can be found on the CIC website.
The next presenter was Judy Ford from the University of California, San Francisco. She presented her project, Event-related brain potentials reveal neurological slowing following COVID-19 infection. This project was funded by the U.S. Department of Veterans Affairs.
Dr. Ford and her team evaluated EEG-based event related protocols (ERPs) to learn about neural slowing following COVID-19. They used objective measures of cognitive function, the P300, which measures the time between the release of a stimulus and an individual’s response to that stimulus. Early P300 responses are typically between 300 and 400 milliseconds while late P300 responses are generally between 380 and 440 milliseconds in latency.
Fox’s research team used this P300 metric to identify respondents’ response times as either passive (“novelty p300”) or active (“target p300”). Test subjects were grouped by their ERP results and post-COVID cognitive functioning self-evaluations. The subjects’ responses were compared to ERPs collected from a pre-pandemic control group. Ford and her team found that both COVID groups , namely those who felt they experienced cognitive decline following infection as well as those who felt they did not experience cognitive decline, exhibited delayed reaction times to auditory stimuli relative to the pre-pandemic control group. This is an ongoing research project, and Ford’s team plans to further examine the relationship between COVID, brain structure, and brain function.
A video of Ford's presentation can be found on the CIC website.
Next, Judy Fox from the University of Virginia presented her research, Expeditions: Collaborative Research: Global Pervasive Computational Epidemiology. This project was funded by the NSF Division of Computing and Communication Foundations (CCF).
Fox’s research interprets county-level COVID-19 infections in the U.S between February 2020 and May 2022. Her team applied the Transformer AI model, a type of deep learning model used by LLMs (Large Language Models), to a dataset developed from static co-variant and dynamic data (e.g. cases or deaths). With this model, they were able to use data inputs from the previous 13 days to predict infection rates for the next 15 days. The researchers’ goal was to use Interpretable AI to identify which counties were at most risk and which communities were most vulnerable to infection. They measured the correlation between the AI county level data and cumulative cases from the CDC of over 3,000 U.S. counties, and concluded that they can interpret the AI model by capturing the self-attention weights at the county level. The model has done well at predicting future waves of COVID cases and Fox’s team believe this can reliably be used to reduce viral transmissions in the future. Their research shows that early interventions can significantly limit infections, underscoring the importance of this research.
A video of Fox's presentation can be found on the CIC website.
Next, Anna Johnson from Georgetown University presented her research, Succumbing, Surviving, and Thriving: The Development of Low-Income Students in the Long Shadow of COVID-19. This project was funded by the NIH National Institute of Mental Health.
Johnson’s research strives to identify the impact of COVID on early childhood development and education. Her team partnered with the largest school district in Oklahoma, the Tulsa Public Schools, as well as community programs that provide for low-income individuals to conduct the Tulsa School Experiences and Early Development Study. This study began in 2016 with a survey of approximately 1,300 low-income three year olds enrolled in Tulsa Public School programs. The researchers followed approximately 1,000 of those students’ progress through 2023, when they began fourth grade. The aim of the original research was to identify the impact of pre-school on low-income students’ long-term academic success. However, the pandemic shifted the focus of this research project. Johnson and her team began to explore the effects of the pandemic on students, including those whose families struggled to maintain medical and financial stability. The study received additional funding and will now be able to follow these students through tenth grade, collecting long-term data on students’ childhood education experiences and the impact of COVID.
A video of Johnson's presentation can be found on the CIC website.
Next, Jayavanth Shenoy from Onai presented his research, Scaled Medical Records Analysis. This project was funded by the NSF Division of Translational Impacts (TI).
The focus of this research at Onai was the scalability, security, and interoperability of medical records. Currently, patients’ medical records are siloed within the systems maintained by their attending physician’s hospital. When cross-hospital research is conducted, the data housed in these records is often transcribed manually, a time-intensive process which also increases the risk of incorrectly coded data. The goal of Shenoy’s project was to enable rapid and secure search of patient medical records across a range of healthcare institutions. With these tools, medical researchers and pharmaceutical companies can reach statistically significant populations more quickly and with greater privacy controls. Onai’s tools implement multi-party computation, allowing researchers to jointly compute results without revealing their inputs to each other or anyone else. Their AI models are also being used by the NIH to support privacy-preserving drug discovery.
A video of Shenoy's presentation can be found on the CIC website.
The webinar was rounded out with a presentation from Kazushige Yokoyama from SUNY Geneseo. He presented his research, Investigating Reversible Aggregation of SARS-CoV-2 Spike Protein-Coated Gold Colloid. This project was funded by the NSF Division of Chemistry.
Yokoyama’s research focuses on the spike protein of the SARS-CoV-2 virus, and in particular, how gold nanoparticles can be used to study the spike protein. Yokoyama discovered that COVID spike proteins produce fibers (amyloidogenesis) when interacting with gold nanoparticles in acidic environments. This is not unlike the amyloid beta proteins which are developed in progressive Alzheimer’s disease. Depending on the pH of the conditions the spike proteins are introduced to, they connect to gold in different ways thereby producing aggregates, folding and unfolding into different shapes. This helps researchers better understand the spike proteins overall. Yokoyama found that the mobility spike protein aggregates can be impacted by the addition of ACE2, an insight which may be crucial in future virology studies.
A video of Yokoyama's presentation can be found on the CIC website.
Following the presentation, Florence Hudson, Lauren Close, and Emily Rothenberg hosted a Q&A session where the audience was able to speak directly with researchers to have their questions answered and engage in enriching discussions.
A recording of this event is available on the Northeast Big Data Innovation Hub’s YouTube Channel and the COVID Information Commons website. The COVID Information Commons is an NSF-funded project brought to you by the Big Data Innovation Hubs, led by the Northeast Big Data Innovation Hub at Columbia University.
We look forward to welcoming you to the next CIC Lightning Talks webinar! Please sign up for the CIC newsletter to be informed of future CIC events.