Introduction to the Special Issue on Understanding the Spread of COVID-19, Part 2

Welcome to the second of two volumes of the ACM Transactions on Spatial Algorithms and Systems’s special issue on “Understanding the Spread of COVID-19” .This volume presents five papers that focus on data-driven models to predict the spread of COVID-19 and to leverage these predictions for effective testing and intervention strategies.

Some of the research directions and articles presented in this and the previous volume have previously been discussed in the ACM SIGSPATIAL Newsletter on the same topic of “Understanding the Spread of COVID-19” [27, 28]. These newsletters included work on COVID-19 data sources [21], mapping mobility changes during COVID-19 [11], COVID-19 cluster detection [12, 15], epidemic simulation [10], contact tracing [17, 26], and spread forecasting [6, 16]. To discuss results and bring researchers together, the “1st ACM SIGSPATIAL International Workshop on Modeling and Understanding the Spread of COVID-19” [5] has been co-located with ACM SIGSPATIAL’20 and the “2nd ACM SIGSPATIAL International Workshop on Spatial Computing for Epidemiology” [4] has been co-located with ACM SIGSPATIAL’21. These interdisciplinary workshops featured keynotes by experts in Epidemiology, Public Health, and Biostatistics and presented eleven more papers on related topics including infection risk estimation [2, 14], disease monitoring [1], epidemic simulation [20, 23, 24], impact of COVID-19 on education [25], COVID-19 cluster detection [3], spatiotemporal analysis [7, 9], and spatiotemporal visualization [22].

To provide a forum to publish research results which may be inspired from the discussions had in aforementioned newsletter articles and workshops, this special issue was announced in January 2021 with a submission deadline on May 31st, 2021. We received 28 submissions, out of which 13 manuscripts have been accepted for publication. Out of the 13 accepted manuscripts, five manuscripts were accepted after one round of revision and eight manuscripts were accepted after two rounds of revision. Out of the 15 rejected manuscripts, 11 were rejected in the first round and four were rejected after a revision. We want to cordially thank the many reviewers around the world for their diligent work which has helped the authors to substantially improve their manuscripts.

The first paper titled “Quantifying the Effects of Contact Tracing, Testing, and Containment Measures in the Presence of Infection Hotspots” by Lorch et al. introduces a temporal point process modeling framework that specifically represents visits to the sites where individuals get in contact and infect each other. This modeling framework allows to explicity model the visits of individuals and sites and allows quantifying the effects of contact tracing, testing, and containment measures in the presence of infection hotspots. Using an efficient sampling algorithm, the authors demonstrate how to estimate the transmission rate of infectious individuals at the sites they visit and in their households using Bayesian optimization and longitudinal case data.

In the second paper titled “Evaluating the Utility of High-Resolution Proximity Metrics in Predicting the Spread of COVID-19”, Mehrab et al. use high-resolution cellphone trace data to extract graph-based proximity metrics which are used to study COVID-19 epidemic spread in 50 land grant university counties in the US. The authors find that, while mobility plays a significant role, the contribution is heterogeneous across the counties. The authors evaluate the utility of the proposed metrics for the prediction case surges.

The third paper titled “Effect of Migrant Labourer Inflow on the Early Spread of COVID-19 in Odisha: A Case Study” by Behera et al. presents a case-study of the effect of migrant labourer inflow on the early spread of COVID-19 in Odisha, India. Authors study the problem of predicting the number of people becoming infected primarily due to reverse-migration. The authors mapped this prediction problem to the Sequential Probability Ratio Test (SPRT) of Abraham Wald. Results show that predictions were highly accurate when compared with real data.

The fourth paper titled “Modeling the Geospatial Evolution of COVID-19 Using Spatio-Temporal Convolutional Sequence-to-Sequence Neural Networks” by Cardoso et al. proposes an approach to model the geospatial evolution of COVID-19 using spatio-temporal convolutional sequence-to-sequence neural networks. To estimate the incidence rate across space for the use-case of Portugal, the authors propose a new convolutional sequence-to-sequence neural network model based on the STConvS2S architecture. Experimental results using real case data show that the proposed model yields better estimations than numerous baseline models.

The final paper titled “SIRTEM: Spatially Informed Rapid Testing for Epidemic Modeling and Response to COVID-19” by Azad et al. proposes an epidemic model that builds on the SEIR model, but is significantly extended to reflect testing, quarantine, and hospitalizations. The model is designed to apply different testing rates for the symptomatic and asymptomatic individuals; for instance, a higher testing rate for the symptomatic individuals. Authors develop an optimization approach to help identify the best possible testing strategy, taking into account daily testing capacities and hospitals physical bed limitations, testing, hospitalization, and quarantine costs. The developed optimization model incorporates realistic spatially based constraints, such as testing capacity and hospital bed limitation.

Enabled by large datasets of human mobility [13] and COVID-19 cases [8], this special issue presents many solutions towards understanding and predicting the spread of an COVID-19. A main open question is whether these solutions can be used to prevent or mitigate future epidemics and pandemics. How well can the patterns of disease spread learned from COVID-19 data be generalized or transferred to other infectious diseases? Answering this question will require close collaboration with sociologists who are able to explain human behavior (such as social distancing behavior or vaccine uptake), epidemiologists to understand the ecology of infectious diseases, and experts in policy to help devise policies to mitigate diseases spread and to effectively communicate such policies to the public. This need to include experts in the mobility data science loop has been described in a recent Dagstuhl report on Mobility Data Science [18].

Another important future direction is not only to include domain experts, but also include decision makers such as health departments. For the United States, this inclusion can be achieved by collaborating with the Department of Health Services in each state or presenting our research results at the Preparedness Summit¹ which is organized by National Association of County and City Health Officials (NACCHO), which represents the 2,800 local public health departments in the U.S. The Preparedness Summit allows to submit abstracts for presentation. While these abstracts are non-archival, the broad impact by sharing results and techniques broadly with decision-makers is extremely valuable.

Another open problem is the access to human mobility, close contacts, and infectious disease data. While this special issue has made an important step forward by making available a dataset of billions of microblogs relating to COVID-19, other datasets are often restricted in access, making it difficult to reproduce research results. Collaboration with data owners is required to improve data curation, sharing, and quality control. In addition, the privacy and ethics issues require much attention especially in public health usage. This aspect has been explored in a recent TSAS Special Issue on Contact Tracing [19]. Additional discussions on privacy and ethics issue beyond contact tracing are needed to ensure that the broad public does not perceive our research on data-driven understanding of the spread of infectious diseases as a threat.