Aggregated mobility data could help fight COVID-19

Buckee, Caroline O.; Balsari, Satchit; Chan, Jennifer; Crosas, Mercè; Dominici, Francesca; Gasser, Urs; Grad, Yonatan H.; Grenfell, Bryan T.; Halloran, M. Elizabeth; Kraemer, Moritz U. G.; Lipsitch, Marc; Metcalf, C. Jessica E.; Meyers, Lauren Ancel; Perkins, T. Alex; Santillana, Mauricio; Scarpino, Samuel V.; Viboud, Cécile; Wesolowski, Amy; Schroeder, Andrew

doi:10.1126/science.abb8021

Cited by 401 publications

(377 citation statements)

References 4 publications

Supporting

Mentioning

353

Contrasting

Unclassified

Order By: Relevance

“…The resulting distribution approximated a Gaussian function. To reduce outlier influence, we constructed a five-level inverted stratification of m50_index , where the highest quintile (5) represented the greatest reduction in mobility since baseline, interpreted as the highest 20% of counties in terms of social distancing intensity. Category boundaries for m50_index by quintile were: lowest mobility change (1) +193% to -45·8%, (2) -46·0% to -55·4%, (3) -55·5% to -62·3%, (4) 62·4% to 74·8%, and highest (5) -75·0% to -100%.…”

Section: Variable Constructionmentioning

confidence: 99%

“…Aggregate movement data from mobile phones have emerged as a potential tool for understanding viral transmission, evaluating interventions, and monitoring compliance with countermeasures. 5 These data could also be used to assess if pandemic response is poised to exacerbate underlying inequities in health. Location tracking includes global positioning system (GPS) data generated from mobile devices (smartphones, tablets, wearable activity trackers, in-vehicle navigation systems, etc.).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Quantifying the social distancing privilege gap: a longitudinal study of smartphone movement

Dasgupta

Lazard

et al. 2020

Preprint

View full text Add to dashboard Cite

Background: In response to the coronavirus pandemic, social distancing became a widely deployed countermeasure in March 2020. We examined whether healthier and wealthier places more successfully implemented social distancing. Methods: Mobile device location data were used to quantify declines in movement by county (n=2,633) in the United States of America, comparing April 15-17 (n=65,544,268 traces) to baseline of February 17 - March 7. Negative binomial regression was used to estimate gradients of privilege across eleven healthcare and economic indicators, adjusting for rurality and stay-at-home mandates. External validation used separate venue-specific data from Google Location Services. Findings: Counties without stay-at-home orders showed a mobility decline of -52.3% (95% CI: -50.3%, -54.3%), slightly less than the decline in mandated areas (-60.8%; 95% CI: -60.0%, -61.6%). Strong linear gradients in privilege were observed. After adjusting for rurality and stay-at-home orders, counties in the highest quintile of social distancing mobility restriction had: 52% less uninsured, 47% more primary care providers, 29% more exercise space, 27% less food insecurity, 26% less child poverty, 17% higher incomes, 14% less overcrowding, 9.6% more racial segregation, 8.2% less youth, 7.4% more elderly, and 6.2% less influenza vaccination, compared to least social distancing areas. Interpretation: Healthier and wealthier counties displayed a social distancing privilege gap, measured via smartphone mobility change. Structural inequities in this key countermeasure will influence immunity, and disease incidence and mortality.

show abstract

Section: Variable Constructionmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Quantifying the social distancing privilege gap: a longitudinal study of smartphone movement

Dasgupta

Lazard

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Since then, the mathematical description of infectious 8 diseases continues to draw significant attention from researchers and practitioners in 9 governments and health agencies alike. Even news agencies are now seeking out 10 explanations to models so as to offer advice and clarity to their audiences during the 11 (near-continuous) coverage of the spread of COVID-19 [2]. The prospect of using 12 mathematical models in conjunction with data is succinctly summarized by the Nobel 13 laureate Ronald Ross, whose 1916 abstract [3] enlightens the role of mathematics in 14 epidemiology today.…”

mentioning

confidence: 99%

Global convergence of COVID-19 basic reproduction number and estimation from early-time SIR dynamics

Katul

Mrad

Bonetti

et al. 2020

Preprint

View full text Add to dashboard Cite

The SIR ('susceptible-infectious-recovered') formulation is used to uncover the generic spread mechanisms observed by COVID-19 dynamics globally, especially in the early phases of infectious spread. During this early period, potential controls were not effectively put in place or enforced in many countries. Hence, the early phases of COVID-19 spread in countries where controls were weak offer a unique perspective on the ensemble-behavior of COVID-19 basic reproduction number R o . The work here shows that there is global convergence (i.e. across many nations) to an uncontrolled R o = 4.5 that describes the early time spread of COVID-19. This value is in agreement with independent estimates from other sources reviewed here and adds to the growing consensus that the early estimate of R o = 2.2 adopted by the World Health Organization is low. A reconciliation between power-law and exponential growth predictions is also featured within the confines of the SIR formulation. Implications for evaluating potential control strategies from this uncontrolled R o are briefly discussed in the context of the maximum possible infected fraction of the population (needed for assessing health care capacity) and mortality (especially in the USA given diverging projections). Model results indicate that if intervention measures still result in R o > 2.7 within 49 days after first infection, intervention is unlikely to be effective in general for COVID-19. Current optimistic projections place mortality figures in the USA in the range of 100,000 fatalities. For fatalities to be confined to 100,000 requires a reduction in R o from 4.5 to 2.7 within 17 days of first infection assuming a mortality rate of 3.4%. : medRxiv preprint in epidemiology. This dispute moved inoculation from the domain of philosophy, 4 religion, and disjointed trials plagued by high uncertainty into a debate about 5 mathematical models -put forth by Daniel Bernoulli (in 1766) and Jean-Baptiste le 6 Rond D'Alembert (in 1761), both dealing with competing risks of death and 7 interpretation of trials [1]. Since then, the mathematical description of infectious 8 diseases continues to draw significant attention from researchers and practitioners in 9 governments and health agencies alike. Even news agencies are now seeking out 10 explanations to models so as to offer advice and clarity to their audiences during the 11 (near-continuous) coverage of the spread of COVID-19 [2]. The prospect of using 12 mathematical models in conjunction with data is succinctly summarized by the Nobel 13 laureate Ronald Ross, whose 1916 abstract [3] enlightens the role of mathematics in 14 epidemiology today. A quotation from this abstract below, which foreshadows the 15 requirements and challenges for mathematical models to describe emerging epidemics 16 such as 5], needs no further elaboration: 17

show abstract

“…active infections at 8 months exceeds the ten thousand case limit), and the outbreak is neither contained nor uncontrolled within 8 months (i.e., fewer than ten thousand active infections at eight months but the outbreak is ongoing). mobility patterns outside of those contexts, using eg., social network data 45 .…”

Section: Discussionmentioning

confidence: 99%

“…The interplay between these sub-community-scale factors and the mobility model should constrain the relative fraction of intra-community transmission with greater confidence, allowing for stronger statements on the efficacy of community-level intervention measures. We are also exploring the use of social network data 45 for improving the interaction model (in particular, mobility patterns outside of work, school and household contexts).…”

Section: Code and Ongoing Workmentioning

confidence: 99%

Containing Covid-19 outbreaks with spatially targeted short-term lockdowns and mass-testing

Alsing

2020

Preprint

View full text Add to dashboard Cite

We assess the efficacy of spatially targeted lockdown or mass-testing and case-isolation in individual communities, as a complement to contact-tracing and social-distancing, for containing SARS-CoV-2 outbreaks. Using the UK as a case study, we construct a stochastic branching process model for the virus transmission, embedded on a network interaction model encoding mobility patterns in the UK. The network model is based on commuter data from the 2011 census, a catchment area model for schools, and a phenomenological model for mobility and interactions outside of work, school, and the home. We show that for outbreak scenarios where contact-tracing and moderate social distancing alone provide suppression but do not contain the spread, targeted lockdowns or mass-testing interventions at the level of individual communities (with just a few thousand inhabitants) can be effective at containing outbreaks. For spatially targeted mass-testing, a moderate increase in testing capacity would be required (typically < 40000 additional tests per day), while for local lockdowns we find that only a small fraction (typically < 0.1%) of the population needs to be locked down at any one time (assuming that one third of transmission occurs in the home, at work or school, and out in the wider community respectively). The efficacy of spatially targeted interventions is contingent on an appreciable fraction of transmission events occurring within (relative to across) communities. Confirming the efficacy of community-level interventions therefore calls for detailed investigation of spatial transmission patterns for SARS-CoV-2, accounting for sub-community-scale transmission dynamics, and changes in mobility patterns due to the presence of other containment measures (such as social distancing and travel restrictions).Disclaimer: We stress that this is a working paper where results are preliminary and subject to change. In particular, we note that the efficacy of spatially targeted interventions are sensitive to the relative proportions of intra-versus inter-community transmission (for a given definition of community boundaries), which in turn is sensitive to the assumptions about the transmission dynamics across different contexts. Whilst the assumptions made here about transmission across contexts are motivated, we are currently updating our model to make the estimated inter-and intra-community transmission rates as robust as possible, as well as running a comprehensive suite of sensitivity tests and different outbreak scenarios.

show abstract

Aggregated mobility data could help fight COVID-19

Cited by 401 publications

References 4 publications

Quantifying the social distancing privilege gap: a longitudinal study of smartphone movement

Quantifying the social distancing privilege gap: a longitudinal study of smartphone movement

Global convergence of COVID-19 basic reproduction number and estimation from early-time SIR dynamics

Containing Covid-19 outbreaks with spatially targeted short-term lockdowns and mass-testing

Contact Info

Product

Resources

About