Contributed Talks – Session I
Presenter: Paul Dempsey, Research Scientist, Dazult
Title: “Modifying SEIR models to explore the role of average household size in the era of lockdowns and social distancing”
Co-Author: James Merrick
Abstract: Compartmental models have long been used to study the dynamics of infectious diseases without requiring the intensive computational power of more detailed simulations. The well mixing assumption on which they are based is typically valid when people are meeting a significant number of other people each day. However, in the Covid-19 era of lockdowns and social distancing, the number of non-household contacts has dropped significantly. Standard SEIR models cannot produce the expected result of an idealised lockdown (i.e. no non household contacts), where the final number of infected is less than the average household size times the number of infected and exposed at the time of the lockdown. To correct this anomaly, we separate the household and non-household contributions to the total cases and apply a carrying capacity to the household acquired cases. We illustrate the application of the model to two countries with completely different approaches to managing the SARS-CoV-2 pandemic, New Zealand and Sweden. The sharp drop in cases in New Zealand following their lockdown can be well explained with a carrying capacity model (Figure 1), while we show that the Swedish approach could be extremely risky for countries with higher average household sizes (Figure 2).
This initial study was undertaken during the first wave of the SARS-CoV-2 pandemic in Spring 2020. We conclude the paper by reflecting on what has been learned since on this topic, the impact of new variants, and the influence of data analysis and modelling on Irish and European Covid-19 policy discussions through the Irish Independent Scientific Advisory Group (ISAG).
Presenter: Stefan Engblom, Professor, Uppsala University
Title: “Bayesian prediction of COVID-19 spread for informed decision making: Practical experiences from Uppsala”
Co-Authors: Robin Eriksson, Hakan Runvik, and Alexander Medvedev
Abstract: Starting from October 2020, the cross-disciplinary project CRUSH Covid is run at Uppsala University, Sweden, to support the authorities of Uppsala County in decision making regarding resource management to mitigate the effects of the pandemic. Conveying objective and timely information to the public is another important role of the project. The project team publishes weekly reports on the COVID-19 status in the county, based on healthcare system information, sewage water analysis, and symptoms monitoring smartphone application data. Mathematical modeling and prediction of COVID-19 spread is part of this effort. The underlying model is a discrete time-invariant linear system of a compartmental structure. The parameters of the model are obtained by Bayesian and system identification techniques. The dynamical states are estimated by a Kalman filter from healthcare data. Estimates and predictions according to the adopted approach are presented up until February, 2021.
Presenter: Gökçe Dayanıklı , PhD Student, Princeton University
Title: “Optimal incentives to mitigate epidemics: A Stackelberg mean field game approach”
Co-Authors: Alexander Aurell, René Carmona, Mathieu Laurière
Abstract: Motivated by models of epidemic control in large populations, we consider a Stackelberg mean field game model between a principal and a mean field of agents evolving on a finite state space. The agents play a non-cooperative game in which they can control their transition rates between states to minimize an individual cost. The principal can influence the resulting Nash equilibrium through incentives so as to optimize its own objective. We analyze this game using a probabilistic approach. We then propose an application to an epidemic model of SIR type in which the agents control their interaction rate and the principal is a regulator acting with non pharmaceutical interventions. To compute the solutions, we propose an innovative numerical approach based on Monte Carlo simulations and machine learning tools for stochastic optimization. We conclude with numerical experiments by illustrating the impact of the agents’ and the regulator’s optimal decisions in two models: a basic SIR model with semi-explicit solutions and in a complex model which incorporates more states.
Presenter: Subhonmesh Bose, Assistant Professor, University of Illinois at Urbana-Champaign
Title: “Modeling and Managing the Spread of COVID-19”
Co-Authors: Ujjal Mukherjee, Sebastian Souyris, Anton Ivanov, Sridhar Seshadri, Yuqian Xu, Padmavati Sridhar and Ronald Watkins
Abstract: Testing, lock-down, and preventative measures such as mask-wearing define non-medical interventions that can mitigate the spread of SARS-CoV-2, the virus behind the COVID-19 pandemic. Testing is a targeted instrument that permits the isolation of individuals who test positive, thereby limiting these individuals’ ability to spread the disease further. Lock-down is a blunt instrument that limits the mobility of all people towards the same goal. Preventive measures such as mask-wearing provide mechanisms to reduce transmission potency from each interaction between infected and susceptible peoples. In our works [BSI+21, MBI+21], we analyze necessary non-medical interventions to contain the spread of COVID-19 through a combination of analytical modeling, data analysis, and numerical simulations.
Epidemic processes within populations are classically studied through compartmental models
that track the sizes of population segments that are susceptible, infected, recovered/deceased, etc.,
e.g., see [KM91]. A distinguishing feature of COVID-19 is its asymptomatic transmission that is not captured in classical models. Our first contribution is a parsimonious SUPR (susceptible-untested-positive-removed) compartmental model that captures the role of asymptomatic transmission, various types of testing, and imposed lock-down measures. Figure 1 presents this model. In Section 1, we describe our key use of this model in analyzing COVID mitigation strategies for a large population. In Section 2, we describe our work on analyzing the same in the setting of an educational institution (e.g., a university with 50K staff/students).
Presenter: Arnab Sarker, PhD Student, MIT
Title: “Forecasting the COVID-19 Pandemic with Mixtures”
Co-Authors: Ali Jadbabaie, Devavrat Shah
Abstract: Throughout the COVID-19 pandemic, there has been a need to understand what comes next. Policy makers, hospitals, and concerned citizens wish to know the state of the pandemic in the next day, week, month, or year. In trying to understand the future of the pandemic, many teams have offered approaches to prediction . These approaches tend to be either mechanistic or non-mechanistic; mechanistic approaches will model the specific transmission parameters of disease, whereas non-mechanistic models will use statistical methods and assume that relevant time series will follow a certain shape . In this work, we focus on a non-mechanistic approach which, upon further examination, has some benefits of the mechanistic approaches. Namely, we assume that the observed time series of either cases or deaths has the form
From a theoretical standpoint, this assumption admits a simple algorithm for learning time series as a mixture of Gaussian curves. Further, such a time-series can be seen as arising from a simple stochastic process on a graph, allowing for meaningful interpretation of parameters. This interpretation can be validated with mobility data, and shows that policy makers can use this curve fitting approach to better understand the progress of the pandemic. From an empirical perspective, we have found that our prediction error is relatively small compared to other models when the number of mixtures r is apparent from the data.
Contributed Talks – Session II
Presenter: X. Flora Meng, PhD Student, MIT
Title: “Impacts of COVID-19 interventions: Health, economics, and inequality”
Co-Authors: Dalton J. Jones, Roberto Rigobon, Munther A. Dahleh
Impact Statement: Coronavirus disease 2019 (COVID-19) is exacerbating inequalities in the US. It is crucial that government interventions address this issue. Our analysis of US data, including income, race, overcrowding, and unemployment, reveals socioeconomic disparities in deaths from COVID-19 and the resulting despair. We build an agent-based model to assess the effects of various nonpharmaceutical interventions (NPIs) on inequality, taking into account factors such as socioeconomic status, age-dependent risks, household transmission, asymptomatic transmission, and hospital capacity. Our simulation produces results that are similar to our observations from US data. First, the trade-off between COVID-19 deaths and deaths of despair, dependent on the lockdown level, only exists in the socioeconomically disadvantaged population. Second, household overcrowding is a strong predictor of the infection rate. Our model also yields new insights that were missing due to data challenges. While subsidization narrows the socioeconomic gap in deaths of despair, the combination of testing and contact tracing alone is effective at reducing disparities in both types of death. Despite the abundance of studies that rely on data analysis or simulation, work that corroborates these two methods is scarce yet essential for policy modeling and evaluation. Our work fills this gap by elucidating the differential causal effects of NPIs on different communities and validating the results with data. Our work has stimulated discussions on policy responses to COVID-19 such as at the C3.ai Symposium on Digital Transformation Science. We have focused on the US, but our approach and results can be extended to other regions in the world.
Presenter: Le Xie, Associate Professor, Texas A&M University
Title: “A Cross-Domain Approach to Analyzing the Short-Run Impact of COVID-19 on the U.S. Electricity Sector”
Co-Authors: Guangchun Ruan, Dongqi Wu, Xiangtian Zheng, Haiwang Zhong, Chongqing Kang, Munther A. Dahleh, S. Sivaranjani
Abstract: As the lifeblood of civil society, the electricity sector is undergoing highly volatile changes due to COVID-19. As the U.S. begins to gradually resume economic activity, it is imperative for policymakers and power system operators to take a scientific approach to understanding and predicting the impact on the electricity sector in both macroscopic (wholesale electricity market) and microscopic (vulnerable groups like low-income households) contexts. In order to assess the macroscopic impact on wholesale markets and the operational challenges posed by these drastic changes to power system operators, we developed a cross-domain, data-driven approach to tracking and quantifying the impact of COVID-19 on the U.S. electricity sector, including (a) a first-of-its-kind open-access data hub COVID-EMDA+ integrating data from across all existing U.S. wholesale electricity markets with public health, mobility, weather, and satellite data, and (b) a cross-domain analysis quantifying the sensitivity of electricity consumption to social distancing and public health policies. Leveraging cross-domain insights from public health and mobility data, we uncover a significant reduction in electricity consumption that is strongly correlated with the number of COVID-19 cases, degree of social distancing, and level of commercial activity.
Presenter: Hope Johnson, Data Fellow, UCLA Law COVID-19 Behind Bars Project
Title: “The Challenges of Creating a Prison Pandemic Database”
Further Information: https://uclacovidbehindbars.org
Abstract: As agencies began reporting COVID‐19 data on infections and deaths inside prisons and jails, the UCLA
Law COVID‐19 Behind Bars Data Project formed to gather data in one place for advocates. What began as a single
spreadsheet quickly snowballed into a complex and multi‐faceted data collection operation involving both qualitative and quantitative data. In this talk, data fellow Hope Johnson will discuss how the unique challenges of criminal justice data have shaped the project’s data collection, validation, and communication efforts. She will also detail how the UCLA Law COVID‐19 dataset has been used by advocates, journalists, and activists to push for greater data transparency and drive urgent public health measures within carceral settings.
Presenter: Nicholas Ashford, Professor, MIT
Title: “Addressing Inequality: The First Step Beyond COVID-19 and Towards Sustainability”
Co-Authors: Ralph P. Hall, Johan Arango-Quiroga, Kyriakos A. Metaxas, Amy L. Showalter
Abstract: The COVID-19 pandemic has impacted billions of lives across the world and has revealed and worsened the social and economic inequalities that have emerged over the past several decades. As governments consider public health and economic strategies to respond to the crisis, it is critical they also address the weaknesses of their economic and social systems that inhibited their ability to respond comprehensively to the pandemic. These same weaknesses have also undermined efforts to advance equality and sustainability. This paper explores over 30 interventions across the following nine categories of change that hold the potential to address inequality, provide all citizens with access to essential goods and services, and advance progress towards sustainability: (1) Income and wealth transfers to facilitate an equitable increase in purchasing power/disposable income; (2) broadening worker and citizen ownership of the means of production and supply of services, allowing corporate profit-taking to be more equitably distributed; (3) changes in the supply of essential goods and services for more citizens; (4) changes in the demand for more sustainable goods and services desired by people; (5) stabilizing and securing employment and the workforce; (6) reducing the disproportionate power of corporations and the very wealthy on the market and political system through the expansion and enforcement of antitrust law such that the dominance of a few firms in critical sectors no longer prevails; (7) government provision of essential goods and services such as education, healthcare, housing, food, and mobility; (8) a reallocation of government spending between military operations and domestic social needs; and (9) suspending or restructuring debt from emerging and developing
countries. Any interventions that focus on growing the economy must also be accompanied by those that offset the resulting compromises to health, safety, and the environment from increasing unsustainable consumption. This paper compares and identifies the interventions that should be considered as an important foundational first step in moving beyond the COVID-19 pandemic and towards sustainability. In this regard, it provides a comprehensive set of strategies that could advance progress towards a component of Sustainable Development Goal (SDG) 10 to reduce inequality within countries. However, the candidate interventions are also contrasted with all 17 SDGs to reveal potential problem areas/tradeoffs that may need careful attention.