The unprecedented global pandemic caused by the spread of the COVID-19 virus comes at a time also marked by historically unparalleled levels of network connectivity, information production, collection and analysis, and increases in community-engaged research across the world. The Information Systems (IS) research community is particularly well-positioned in exploring groundbreaking ways to leverage digital technologies, interdisciplinary approaches, and community resources to develop effective responses to the pandemic and its aftermath. NSF RAPID funded projects in the UMBC IS department include, Deep Learning Models for Early Screening of COVID-19 using CT Images, Influences of the COVID-19 Outbreak on Racial Discrimination, Identity Development and Socialization, and Responding to COVID-19 using High-speed Mesh Wireless Community Internet. Other research efforts, such as USM’s Maryland Institute for Pandemic Preparedness, have initiated to leverage the potential of interdisciplinary efforts in this area.
In recognition of the innovative research being conducted by the IS faculty and with the goal of increasing research visibility and crosscutting collaborations, the IS research committee organized a virtual showcase on Friday, November 13th on Pandemic Preparedness Research. The event consisted of a panel and a series of lightning talks. Below, we include brief summaries of the projects that were shared.
The panel, which was moderated by Dr. Nirmalya Roy, showcased three NSF RAPID-funded projects. After providing an overview of their projects, the panelists discussed the mechanics of applying for RAPID funding, strategies for success during a limited project timeline, and plans on project continuation beyond the grant. Following are brief descriptions of each of the three projects:
Deep Learning Models for Early Screening of COVID-19 using CT Images (Award # 2027628)— Dr. Aryya Gangopadhyay: The rapid spread of COVID-19 has severely impacted the lives of billions of people across the world. Healthcare systems are strained both in terms of dealing with the large number of cases but also the risk of infection imposed on healthcare workers. This project develops low-cost, effective, and minimal contact early screening tools for detection, treatment, and prevention of the spread of the disease. In order to respond to infectious diseases such as COVID-19 and prevent future, this project proactively builds resources to help the medical community be better prepared in early stages of diseases with pandemic potential. This project develops an understanding of SARS-CoV-2 through an early screening tool to distinguish the recent coronavirus (COVID-19) infections from other respiratory illnesses such as Influenza-A and viral or bacterial pneumonia as well as from patients who have no pulmonary disease. There are two major contributions of the project: (1) generate high-quality Convolutional Neural Networks (CNNs) with 2D and 3D kernels for early detection of COVID-19 infection, and (2) synthesize realistic Computed Tomography (CT) images using Generative Adversarial Networks (GANs) that will be publicly available for research and practice.
Covid Tweet Analysis with Latent Knowledge Models (Award #2024124)— Dr. Shimei Pan: Social media platforms such as Twitter contain an unprecedented amount of user-generated content related to the Covid-19 pandemic. In this research, we demonstrate how social media analytics, especially latent knowledge models trained on a large amount of social media text, can be used to help us understand the evolution of Covid-related public discourse. We also compare the public narratives from multiple regions (e.g., US, China, and the Middle East) via multilingual text analytics.
Responding to COVID-19 using High-speed Mesh Wireless Community Internet (Award # 2030451)— Dr. Foad Hamidi: In this collaborative project, we are responding to COVID-19 by investigating an effective and efficient community-based approach in Baltimore City, Maryland to deploying free, broadband Internet and creating trusted open-access online education, career, and communication resources for low-income populations in the face of large-scale emergencies. This approach builds on existing research on the importance of equitable broadband Internet access and the potential of community-based solutions to bridging the digital divide. Project findings will inform the creation and use of community-led approaches to meet the technical and informational needs of vulnerable populations during and immediately following times of crisis. Specifically, working with Project Waves and the Digital Harbor Foundation (DHF), our research team at UMBC is investigating the creation of a trusted technical infrastructure that leverages local partnerships to provide free or low-cost Internet to communities. It will also inform how to maximize the potential of Internet connectivity to maintain continuity of education and employment activities and reduce social isolation among low-income populations. A video outlining the project is available on youtube.
The panel was followed by three lightning talks on pandemic-related research:
Exploratory analysis of covid-19 tweets using topic modeling, umap, and digraphs by Dr. Sanjay Purushotham. This paper illustrates five different techniques to assess the distinctiveness of topics, key terms and features, speed of information dissemination, and network behaviors for Covid19 tweets. First, we use pattern matching and second, topic modeling through Latent Dirichlet Allocation (LDA) to generate twenty different topics that discuss case spread, healthcare workers, and personal protective equipment (PPE). One topic specific to U.S. cases would start to uptick immediately after live White House Coronavirus Task Force briefings, implying that many Twitter users are paying attention to government announcements. We contribute machine learning methods not previously reported in the Covid19 Twitter literature. This includes our third method, Uniform Manifold Approximation and Projection (UMAP), that identifies unique clustering-behavior of distinct topics to improve our understanding of important themes in the corpus and help assess the quality of generated topics. Fourth, we calculated retweeting times to understand how fast information about Covid19 propagates on Twitter. Our analysis indicates that the median retweeting time of Covid19 for a sample corpus in March 2020 was 2.87 hours, approximately 50 minutes faster than repostings from Chinese social media about H7N9 in March 2013. Lastly, we sought to understand retweet cascades, by visualizing the connections of users over time from fast to slow retweeting. As the time to retweet increases, the density of connections also increases where in our sample, we found distinct users dominating the attention of Covid19 retweeters. One of the simplest highlights of this analysis is that early-stage descriptive methods like regular expressions can successfully identify high-level themes which were consistently verified as important through every subsequent analysis. A video of the presentation is available on youtube.
A Model for Examining the Impact of the COVID-19 Pandemic on the Undergraduate Student Experience and Designing Equitable Online Supports for Underrepresented Technology Majors by Dr. Carolyn Seaman: The Coronavirus Disease 2019 (COVID-19) pandemic has altered undergraduate education in unprecedented ways, resulting in abrupt campus closures, unplanned transitions to online classes, and virtual student support services. A proposed project submitted to the NSF by UMBC’s CWIT and other collaborators addresses the need for research about the consequences of this new educational context for students and educators to provide knowledge about emergency remote teaching and learning (ERTL; see Hodges, et al., 2020) in STEM. Further, women and underrepresented minority students may be even more vulnerable to academic marginalization, access limitations, and bias in ERTL settings in computing and engineering — two of the STEM fields with the least traction in diversity efforts pre-pandemic. The effects of this disruption in higher education will be felt for many years to come and will require new and relevant responses to maintain a diverse and prepared workforce in computing and engineering.
In the past six months of global pandemic, educational research and journalistic publications have substantiated the impact of COVID-19 and ERTL on the conditions in which students learn; however, we need insight on the discipline-specific implications of this new educational context for students’ curricular, co-curricular, and extra-curricular experiences in computing and engineering, as well as their perceptions of themselves and the educational climate. The results of this project will provide insight into solutions to challenges students face across critical transitions including from high school to university, from community college to university, and pursuit of post-baccalaureate plans in the midst of ERTL. In addition to identifying and meeting the needs of students who have experienced ERTL, the project will also identify successful student supports and pedagogical strategies for inclusive online learning in computing and engineering education that are endorsed by faculty as well as underrepresented students.
Discovery of Robust Distributions of COVID-19 Spread by Dr. Vandana Janeja: We are currently living in a world receiving a fire hose of COVID-19 data. At the same time many executives on the front lines are expressing that they are “flying blind” and “do not know where the outbreak” will hit in their own local jurisdictions. Others have also raised concerns on not being able to tell if they are at a peak or plateau or whether they can reopen with worries of a resurgence of cases, especially in the absence of widespread COVID-19 tests. Some jurisdictions have attempted to report cases by zip code to assess the correlations of cases and deaths by demographic features to grasp the impact on minorities. Models have been proposed to predict the COVID-19 cases and deaths, visualize spatio-temporal data for outbreaks, study the impact of environmental features. These types of analysis and visualizations track the current trends and predict the numbers of cases and deaths based on the trend observed so far. As a result, they are dependent on several days of case and mortality data. Each location has underlying heterogeneity that comes into play when the outbreak progresses over a period of time and is exacerbated by underlying socio-economic, health, and other features at the location. These underlying features need to be accounted for, to study the level of impact of the disease at any given time and place. This can help predict the trend and resource needs for a region, identify locations that could be hard hit if they resemble other high-intensity regions. On the other hand, two locations that have very similar underlying features could have different outcomes, this is critical to identify so mitigation strategies can be studied which potentially lead to different intensities of the spread. This project proposes a temporal mining methodology to discover distributions of the disease spread which can be used to compare spreads across such heterogeneous locations and time and facilitate comparative analysis of the disease spread.
As can be seen from the summaries in this article, there was a wide range of research approaches and techniques that were represented in the research projects. A thread that ran through many of the projects was the use of interdisciplinary collaborative approaches to tackle complex and multifaceted issues related to pandemic preparedness. We would like to thank all participants and audience members who attended this event.