Early risk-assessment of pathogen genomic variants emergence (preprint)

Accurate, reliable, and timely estimates of pathogen variant risk are essential for informing effective public health responses to infectious diseases. Despite decades of use for influenza vaccine strain selection and PCR-based molecular diagnostics, data on pathogen variant prevalence and growth advantage has only risen to its current prominence during the SARS-CoV-2 pandemic. However, such data are still often sparse: novel variants are initially rare or a region has limited sequencing. To ensure real-time estimates of risk are available in these types of data-sparse conditions, we develop a hierarchical modeling approach that estimates variant fitness advantage and prevalence by pooling data across geographic regions. We apply this method to estimate SARS-CoV-2 variant dynamics at the country-level and assess its stability with retrospective validation. Our results show that more stable and robust estimates can be obtained even when sequencing data are sparse, as compared to established, single-country estimation approaches. We discuss how this method can inform risk assessment of novel emerging variants and provide situational awareness on currently circulating variants, for a range of pathogens and use-cases.

Characterizing spatiotemporal trends in self-reported mask-wearing behavior in the United States (preprint)

Although face mask-wearing has been adopted throughout the U.S. to prevent the spread of COVID-19, reliable spatial estimates of mask-wearing through different phases of the pandemic do not yet exist. Using 8+ million survey responses, survey raking, and debiasing with ground-truth data on a different mitigation behavior, we generate fine-scale spatiotemporal estimates of mask-wearing across the U.S. from September 2020 to May 2021. We find that county-level masking behavior is spatially heterogeneous along an urbanrural gradient and moderately temporally heterogeneous. Because these survey data could be prone to social desirability and non-response biases, we evaluate whether a question about community mask-wearing could be a less biased alternative and find support for this social sensing approach to behavioral surveillance. Our work highlights the need to characterize public health behaviors at fine spatiotemporal scales to capture heterogeneities driving outbreak trajectories, and the role of behavioral big data to inform public health efforts.

Disentangling the rhythms of human activity in the built environment for airborne transmission risk (preprint)

Since the outset of the COVID-19 pandemic, substantial public attention has focused on the role of seasonality in suppressing transmission. Misconceptions have relied on seasonal mediation of respiratory diseases driven solely by environmental variables. However, seasonality is expected to be driven by host social behavior, particularly in highly susceptible populations. A key gap in understanding the role of social behavior in respiratory disease seasonality is our incomplete understanding of the seasonality of indoor human activity. We leverage a novel data stream on human mobility to characterize activity in indoor versus outdoor environments in the United States. We use a mobile app-based location dataset encompassing over 5 million locations nationally. We classify locations as primarily indoor (e.g. stores, offices) or outdoor (e.g. playgrounds, farmers markets), disentangling location-specific visitor counts into indoor and outdoor, to arrive at a fine-scale measure of indoor to outdoor human activity across time and space. We find the proportion of indoor to outdoor activity during a baseline year is seasonal, peaking in winter months. The measure displays a latitudinal gradient with stronger seasonality at northern latitudes and an additional summer peak in southern latitudes. We statistically fit this baseline indoor-outdoor activity measure to inform incorporation of this complex empirical pattern into infectious disease dynamic models. However, we find that the disruption of the COVID-19 pandemic caused these patterns to shift significantly from baseline, and the empirical patterns are necessary to predict spatio-temporal heterogeneity in disease dynamics. Our work empirically characterizes, for the first time, the seasonality of human social behavior at a large-scale with high spatio-temporal resolution, and provides a parsimonious parameterization of seasonal behavior that can be included in infectious disease dynamics models. We provide critical evidence and methods necessary to inform the public health of seasonal and pandemic respiratory pathogens and improve our understanding of the relationship between the physical environment and infection risk in the context of global ecological change.

Characterizing the spatiotemporal heterogeneity of the COVID-19 vaccination landscape (preprint)

It is critical that we maximize vaccination coverage across the United States so that SARS-CoV-2 transmission can be suppressed, and we can sustain the recent reopening of the nation. Maximizing vaccination requires that we track vaccination patterns to measure the progress of the vaccination campaign and target locations that may be undervaccinated. To improve efforts to track and characterize COVID-19 vaccination progress in the United States, we integrate CDC and state-provided vaccination data, identifying and rectifying discrepancies between these data sources. We find that COVID-19 vaccination coverage in the US exhibits significant spatial heterogeneity at the county level and statistically identify spatial clusters of undervaccination, all with foci in the southern US. Vaccination progress at the county level is also variable; many counties stalled in vaccination into June 2021 and few recovered by July, with transmission of the Delta variant rapidly rising. Using a comparison with a mechanistic growth model fitted to our integrated data, we classify vaccination dynamics across time at the county scale. Our findings underline the importance of curating accurate, fine-scale vaccination data and the continued need for widespread vaccination in the US, especially in the wake of the highly transmissible Delta variant.

Ignoring spatial heterogeneity in drivers of SARS-CoV-2 transmission in the US will impede sustained elimination (preprint)

To dissect the transmission dynamics of SARS-CoV-2 in the United States, we integrate parallel streams of high-resolution data on contact, mobility, seasonality, vaccination and seroprevalence within a metapopulation network. We find the COVID-19 pandemic in the US is characterized by a geographically localized mosaic of transmission along an urban-rural gradient, with many outbreaks sustained by between-county transmission. We detect a dynamic tension between the spatial scale of public health interventions and population susceptibility as pre-pandemic contact is resumed. Further, we identify regions rendered particularly at risk from invasion by variants of concern due to spatial connectivity. These findings emphasize the public health importance of accounting for the hierarchy of spatial scales in transmission and the heterogeneous impacts of mobility on the landscape of contagion risk.

Characterizing superspreading of SARS-CoV-2: from mechanism to measurement (preprint)

Superspreading is a ubiquitous feature of SARS-CoV-2 transmission dynamics, with a few primary infectors leading to a large proportion of secondary infections. Despite the superspreading events observed in previous coronavirus outbreaks, the mechanisms behind the phenomenon are still poorly understood. Here, we show that superspreading is largely driven by heterogeneity in contact behavior rather than heterogeneity in susceptibility or infectivity caused by biological factors. We find that highly heterogeneous contact behavior is required to produce the extreme superspreading estimated from recent COVID-19 outbreaks. However, we show that superspreading estimates are noisy and subject to biases in data collection and public health capacity, potentially leading to an overestimation of superspreading. These results suggest that superspreading for COVID-19 is substantial, but less than previously estimated. Our findings highlight the complexity inherent to quantitative measurement of epidemic dynamics and the necessity of robust theory to guide public health intervention.