We will continue our research into the value and impact of using mobility data to understand the effects of Covid-19 and lockdowns. Data for development: the d4d challenge on mobile phone data. Excluding South Korea, we estimate that all policies combined were associated with a decrease in mobility by 81% . So far, more than 1,000 organizations including the CDC are already in the Safegraph data consortium. Tour. medRxiv (2020). Nature 581, 109111 (2020). A tag already exists with the provided branch name. We can get some insights on this from the data that Google presents in its COVID-19 Community Mobility Reports. Appendix Table B.21 shows that for all recreational locations in the SafeGraph data, . There are several scenarios for how this interaction may work: Every country, state, city, or area will have its own dynamic. PubMed (e) Illustrative example of different mobility measures in California. The move came on the heels of a Vice article raising concerns that SafeGraph data could . C.I., J.B., S.A.P., S.H., X.H.T., designed analysis, and interpreted results. In general, sub-national forecasts in China benefit least from mobility data, but forecasts in Italy and the US are substantially improved by including a single measure of mobility for the 21 days prior to the date of the forecast. arXiv preprint arXiv:2004.10172 (2020). Science 368, 395400 (2020). We evaluate this approach using local and regional data from China, France, Italy, South Korea, and the United States, as well as national data from 80 countries around the world. Here, we aim to address this modeling-capacity gap by developing, demonstrating, and testing a simple approach to forecasting the impact of NPIs on infections. Chang, S. et al. In principle, such future forecasts can be used by decision-makers who are able to influence local mobility through policy and/or NPIs, perhaps informed either by a behavioral model or observation. S.A.P. Can you use the model to predict what will happen in the next weeks/months? The policy data was constructed and made available for academic research by Global Policy Lab2,29. The volumes of mobility data being collected now, particularly in cities are huge. SafeGraph data is freely available to researchers, non-profits, and governments through the SafeGraph COVID-19 Data Consortium. 1 and Table S1. doi: 10.4081/gh.2022.1056. For example, in Chicago, the model predicts that 10% of POIs accounted for 85% of infections at POIs. https://www.kaggle.com/jcyzag/covid19-lockdown-dates-by-country. The effect of travel restrictions on the spread of the 2019 novel coronavirus (covid-19) outbreak. We merge the daily country-level observations to construct a longitudinal data sets for the portion of the world we observe. CNN; Martn-Calvo, D., Aleta, A., Pentland, A., Moreno, Y. 50, 801808 (2020). & Weber, M. The Cost of the Covid-19 Crisis: Lockdowns, Macroeconomic Expectations, and Consumer Spending, Technical Report, National Bureau of Economic Research (2020). In the example below, Kexin Mao uses Google Community Mobility Report data to visualize state-by-state movement patterns and compare where people are going and how that deviated from the pre-pandemic norm. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. The effect of human mobility and control measures on the COVID-19 epidemic in China. In lower-resource settings, where use of smartphones is less common, the users who generate mobility data may not be as representative of the total population as in wealthy nations, but prior work suggests that biases in phone ownership may not dramatically bias estimates of overall population mobility41. Report 9: impact of non-pharmaceutical interventions (npis) to reduce covid19 mortality and healthcare demand (2020). This movement is likely correlated with other behaviors and factors that contribute to the spread of the virus, such as low rates of mask-wearing and/or physical distancing. \end{aligned}$$, \(\frac{\Delta infections}{\Delta behavior}\), https://doi.org/10.1038/s41598-021-92892-8. It is designed to enable any individual with access to standard statistical software to produce forecasts of NPI impacts with a level of fidelity that is practical for decision-making in an ongoing crisis. We merge the sub-national NPI, mobility, and epidemiological data based on administrative unit and day to form a single longitudinal (panel) data set for each country. The data from SafeGraph, which says it tracks only users who have "opted in" via mobile . Global Health Action 13, 1816044 (2020). Nat. Available at SSRN (2020). Ilin, C., Annan-Phan, S., Tai, X.H. Full details, including model equations and estimation methods, are provided in Supplementary file 1: AppendixB. https://doi.org/10.7910/DVN/FAEZIO. Transport planners will need to understand the volumes using transport so they can ensure that they can strike a balance between safety and efficiency. Policy (2020). | Find, read and cite all the research you . These private companies provide free aggregated and anonymized information on the movement of users of their online platform (Fig. Sign up for the Nature Briefing newsletter what matters in science, free to your inbox daily. Mobility is represented as daily total number of visits to points of interest (any non-residential place), based on aggregated geolocation data from SafeGraph. Country-level forecasts, which use country-level mobility data from Google, benefit relatively less than sub-national model from including mobility information, in part because baseline forecast errors are smaller. The country-specific analysis is determined by data availability. This is our first blog post about mobility data and Covid-19; future work will focus on what mobility data users need and the barriers they may face. Safegraph is. https://huiyan.baidu.com. A large number of visits to local businesses increased the positivity rate of COVID-19 tests, while a large number of smartphones that stayed at home decreased it. In Supplementary file 1: AppendixC, we disaggregate this effect temporally, and find that the most significant reductions occur during the first eight days after a lockdown (FigureS1c). (2021). However, mobility data bias has received little attention in this predictive context. Nature, 2020. Do you use data on how many people were infected at different types of places? Mobility network models of COVID-19explain inequities and inform reopening. We obtained an IRB exemption for SafeGraph data from Northwestern University. The infection model describes how infections change in association with changes in mobility behavior (\(\frac{\Delta infections}{\Delta behavior}\)). We assume that mobility patterns to all POIs return to pre-pandemic levels, except that hourly visits to each POI are capped so they cannot exceed a fraction of the POIs maximum occupancy. This graph depicts the cumulative COVID-19 case counts predicted by our model under your scenario (red), compared to our model predictions when run with actual mobility data (green), which closely track real case counts (as reported by The New York Times). Funding was also provided by Award 2020-0000000149 from CITRIS and the Banatao Institute at the University of California. Read-in big data in chunks while filtering on only relevant rows (in this case rows pertaining to Austin, TX), Explore connecting to Google Drive to save smaller chunks of data. Facebook Disaster Maps. Mar 16, 2021, 09:01 ET DENVER, March 16, 2021 /PRNewswire/ -- SafeGraph, a data-as-a-service company focused on being the source of truth for data on physical places, announced today a $45. A panel multiple linear regression model is used to estimate the relative association of each category of mobility with each NPI. Using anonymized data provided by apps such as Google Maps, the company has produced a regularly updated dataset that shows how peoples' movements have changed throughout the pandemic. Figure 1. The reduced-form approach presented here can still be applied in such circumstances, but it may be necessary to refit the model based on data that is representative of current conditions. Because SafeGraph and other location providers gather mobile identifiers and precise, time-stamped latitudinal and longitudinal location coordinates, privacy and abortion rights advocates fear that the information could be used to detect when specific people have visited abortion clinics or other sensitive locations, particularly if only a few devices are present in a place at a given time. Zhang, S. X. et al. Association of mobile phone location data indications of travel and stay-at-home . Reopening does not have to be all-or-nothing: strategies like reducing maximum occupancy can enable us to reopen more efficiently by providing a large reduction in infections for a relatively small reduction in visits. Using these estimated changes in mobility, they could then forecast changes in infections using the infection model described abovebut fit to local data. Importantly, the risk to society of fully reopening a category is not equivalent to how risky it is for you, as an individual, to visit a POI in that category now. S. Gao, J. Rao, Y. Kang, et al. Limited data availability has hindered model development and evaluation since the inception of agent-based modeling in the late 1980s [6]. Results are provided at the prefecture (ADM2) and province level (ADM1) in China; the regional (ADM1) level in France; the province (ADM2) and region (ADM1) level in Italy; the province (ADM1) level in South Korea; and the county (ADM2) and state (ADM1) level in the United States. In practice, we estimate a distributed-lag model where the predictor variables are mobility rates in that location for the prior 21 days, and the dependent variable is the daily infection growth rate, constructed as the first-difference of log confirmed infections. S.M. SafeGraph mobility data includes information about foot traffic at over 5 million places across the US based on cell phone records [ 14 ]. Our data records how many people go to points of interest (POIs) like restaurants and grocery stores at every hour, and also records the neighborhoods they come from. They are publicly available at different locations. We would like to speak to users, producers and publishers of mobility data, so if this is you please do get in touch, Course, Members Event, ODI Summit 2022 taster session, Online, Online Course, Workshop, Datopolis: The open data board game [taster session @ the ODI Summit 2022]. The simple model we present here is designed to provide useful information in contexts when more sophisticated process-based models are unavailable, but it should not necessarily displace those models where they are available. Short term prediction of COVID-19 cases. What does your model say about socioeconomic and racial disparities in COVID infection rates? Technical Report, National Bureau of Economic Research (2020). The data includes aggregated and anonymized datasets on social distancing and foot traffic to businesses.". Traffic data such as that produced by TomTom can also be used to compare cities across the country (and the world) to assess the status of lockdowns and comparative movement. Data Ethics Professionals and Facilitators. A doubling in the relative number of stay-at . Coibion, O., Gorodnichenko, Y. This approach is also robust to incomplete rates of COVID-19 testing, uneven patterns of testing across space, and gradual changes in testing over time2see Supplementary file 1: AppendixB.2 for details. The New York Times; Provided by the Springer Nature SharedIt content-sharing initiative. Similarly, for data fitted at a global level (bottom-most plot), for each country and forecast length, the mean is taken over all forecast dates. The California Governor's Office relies on SafeGraph data to develop COVID-19 policies, including risk measurements of specific areas and facilities and enforcement of physical distancing. Google Scholar Crane, M. A., Shermock, K. M., Omer, S. B., et al. However, this does not imply that population mobility itself is the only fundamental cause of transmission. The Centers for Disease Control and . We estimate the reduction in human mobility associated with the deployment of NPIs by linking comprehensive data on policy interventions to mobility data from several different countries at multiple geographic scales. The data from the Apple and CItymapper mobility reports is generated when the user requests directions. Thus, human mobility flows play a crucial role in the spatial spread of the virus; the heterogeneity of mobility patterns and social distancing behavior can largely explain the geographic heterogeneity of transmission ( 3 - 11 ). We distinguish between three different levels of aggregation for administrative regions - denoted ADM2 (the smallest unit), ADM1, ADM0. Our global analysis is conducted using ADM0 data. Safegraph uses footfall data to demonstrate consumer activity, in a similar manner in the US. 11(2), 179195 (2020). What does your model say about the risks of different categories of places, like restaurants or gyms? First, we show that passively collected data on human mobility, which has previously been used to measure NPI compliance20,21,22,23,24,25,26, can also effectively forecast the COVID-19 infection response to NPIs up to 10 days in the future. Davis, used mobility data from SafeGraph, PlaceIQ and Google Mobility from January 2020 to . Figure4 summarizes model performance across all administrative subdivisions of each of the three countries we consider for the forecast analysis (China, Italy, and the United States). (a) Home isolation policy adoption, (b) Change in time spent at home, (c) Infection growth rate, and (d) Total confirmed cases are displayed at the county, state and country level. Markers are country specific-estimates, whiskers show the 95% confidence interval. We are working on analysis with more recent data. The Mobility and Engagement Index created by economists at the Dallas Fed uses geolocation data collected from mobile devices by a company called SafeGraph. Enterprise-level solutions for managing spatial data. For countries in our sample, MPE is 6.35% (5-day) and 15.24% (10-day) accounting for mobility, and 11.46% and 31.12% omitting mobility. Data on mobility measures, COVID-19 infections and home isolation policy adoption. In the USA and Italy, the impact of NPIs on mobility was highly localized, with little evidence of spatial spillover effects (Supplementary file 1: AppendixC - FigureS1a). The general consistency of these magnitudes across countries holds for alternative measures of mobility: using Google data we find that all NPIs combined result in an increase in time spent at home by 28% (se = 2.9), 24% (se = 1.3), and 26% (se = 1.3) in France, Italy, and the US, respectively. Furthermore, identical models that exclude mobility data perform substantially worse, suggesting an important role for mobility data in forecasting. How large that probability is depends on the area of the POI, how long visitors stay there, and how many of the current visitors are infectious. is a Chan Zuckerberg Biohub investigator. Google (2020). In this study, the first independent audit of demographic bias of a smartphone-based mobility dataset used in the response to COVID-19, researchers assessed the validity of SafeGraph data.. Understanding how people are moving is therefore important for government authorities, transport planners and epidemiological researchers as well as others to understand the effects of the pandemic and policy actions. With mobility-based features integrated with the typical load forecasting features, we were able to predict peak electricity demand without using historic load data, which was significantly fluctuated . The combined effects were of similar magnitude in China ( 78%, se = 8%), France ( 88%, se = 27%), Italy ( 85%, se = 12%), and the US ( 69%, se = 6%); no significant change was observed in South Korea, where mobility was not a direct target of NPIs (for example39). For example, a forecast made for the period 4/06/20204/15/2020 for California-Los Angeles on 4/15/2020 without mobility projects 30,716 cases, while the same forecast accounting for mobility would be 12,650 cases, much closer to the 10,496 that was observed. We thank Jeanette Tseng for her role in designing Fig. This data can be a useful indicator for movement. Based on these observed responses, they could forecast infections using our behavior model. J.L. COVID-19 Community Mobility Reports. given that safegraph' samples are highly correlated with the true census populations regarding several socio-economic attributes 51, we aim to infer the short-term population-level dynamic. No. Read more about SafeGraph and the data they are collecting here. Correspondence to If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. Societies and decision-makers around the globe are deploying unprecedented non-pharmaceutical interventions (NPIs) to manage the COVID-19 pandemic. Similar tabulations can be generated by fitting infection models using recent and local data, which would flexibly capture local social, economic, and epidemiological conditions. This material is based upon work supported by the National Science Foundation under Grant IIS-1942702, the Office of Naval Research (Minerva Initiative) under award N00014-17-1-2313, and CITRIS and the Banatao Institute at the University of California under Award 2020-0000000149. Mobile phone data can be used in the coronavirus pandemic to understand the volume of the population moving, to answer cause-and-effect questions on different control mechanisms such as lockdowns, to predict future needs, risks and opportunities and to overall assess the effectiveness of different types of intervention. The data will be useful to make decisions about lifting restrictions and restarting the economy. ToPLAYDatopolis at the ODI Summit, youll need tobuy an ODI Summit 2022 ticketand apply below to secure your place places are limited to 6 players. Rather, it represents a practical and low-cost alternative that may be easily adopted in many contexts when the former is unavailable. The reduced-form model we develop generally performs well when fit to local data, except in China where it cannot account for some key factors that contributed to reductions in transmission. Project (s): GIS data and technology projects for midstream energy company focused on environmental impact and . In cases where complete process-based epidemiological models have been developed for a population and can be deployed for decision-making, the model we develop here could be considered complementary to those models. We provide a brief summary of these data here; full details are provided in Supplementary file 1: AppendixA. Use Git or checkout with SVN using the web URL. A public authority runs a service themselves and collects data about users. Kraemer, M. U. G. et al. We show that publicly available data on human mobilitycollected by Google, Facebook, and other providerscan be used to evaluate the effectiveness of non-pharmaceutical interventions (NPIs) and forecast the spread of COVID-19. There are a number of differences: 1) we study the risk of reopening the entire category, not the risk of one person visiting one of these places; 2) POIs within the same category vary a lot in how risky they are; 3) we study data from the spring, but nowadays many places have modified their levels of mobility and may also be taking additional precautions like mask-wearing. According to a Science Advances article, Control of the pandemic requires control of people including their mobility and other behaviors. Around the world, people of all ages have been confined to their homes following government orders in many countries to only go outside for what were deemed as essential activities. J. At the local (ADM2) level in Italy, the MPE is 1.73% and 13.27% for five and ten days in the future when mobility is accounted for, compared to 45.81% and 167.97% when it is omitted. A key insight from our work is that passively observed measures of aggregate mobility are useful predictors of growth in COVID-19 cases. With the investment, SafeGraph plans to. Similarly, the national emergency declaration was associated with significant mobility reductions in China (- 62.6 %, se = 12.7 %). We use a spatiotemporal agent based model that is informed by Safe Graph Data to improve the accuracy of the model. streams for influenza and other diseases, and pioneered many of the API concepts discussed below. Policymakers everywhere are working to determine the set of restrictions that will effectively contain the spread of COVID-19 without excessively stifling economic activity. J. Tian, H. et al. Interactive simulation frontend produced in collaboration with J.D. Our model predicts that lower income and less white neighborhoods will have higher infection rates, which is consistent with what actually happened during the time period we model. There are two exceptions to this rule, to include select industrial POIs and corporate offices for major organizations. Google Scholar. We conclude by discussing how these models could be used to guide policy decisions at local and regional scales. Hum. 2020. https://doi.org/10.1080/09669582.2020.1758708. (2020). We hypothesize that the approach we develop here might skillfully forecast the spread of other diseases besides COVID-19. If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate. Nat. Thus, different populations have adopted wildly different containment strategies11, and local decision-makers face difficult decisions about when to impose or lift specific interventions in their community. An investigation of transmission control measures during the first 50 days of the covid-19 epidemic in china. 2a). 2b). As with the behavior model, we model the daily growth rate of infections at the local, national, and global scale. These changes are relative to a baseline defined as the median value, for the corresponding day of the week, during Jan 3Feb 6, 2020. J.B. and S.H. Daily mobility measures based on anonymized and aggregated mobile device data were obtained from SafeGraph, Google, and Place IQ. A study published in the journal Nature by Serina Chang and her colleagues in November 2020 used American cell phone data to identify the places where people are most likely to contract SARS-CoV-2, the virus that causes Covid-19. was supported by a Hertz Fellowship. https://www.oecd.org/coronavirus/en/#country-tracker. Our study links information on non-pharmaceutical interventions (NPIs, shown in Fig. We aggregate these data to show trips between and within sub-national units. This approach is not a substitute for more refined epidemiological models. http://www.globalpolicy.science/covid19. Lastly, SafeGraph dataset gives us information on average distance travelled from home by millions of devices across the US36. Safegraph uses footfall data to demonstrate consumer activity, in a similar manner in the US. This means that even stringent occupancy caps can result in relatively small reductions in the total number of visits because they only affect businesses during their most crowded hours, and leave visit patterns during less crowded hours unchanged. Mobile phone data for informing public health actions across the covid-19 pandemic life cycle (2020). In all geographies and at all scales, models with mobility data perform better than models without. The Organisation for Economic Co-operation and Development (2020). Data describing peoples movements from one location to another and the mode of transport used is known as mobility data. We explore how the ever-increasing volumes of mobility data can help monitor lockdown adherence, explain the spread of disease, and assist with transport decision-making. Nature 19 (2020). Community mobility was defined as the percentage of personal mobile devices (e.g., mobile phones, tablets, and watches) leaving home, using publicly accessible data from SafeGraph, a data company that aggregates anonymized location data from mobile devices ( 5 ). However, this data is only indicative of movement. At the sub-national level, we use the NPI dataset compiled by Global Policy Lab2,29. If nothing happens, download GitHub Desktop and try again. We do not recommend using our findings about risky POIs to plan your daily life, because our analysis is designed for policymakers, not individuals (see our answer above to What does your model say about the risks of different categories of places, like restaurants or gyms?). These effects are not modeled explicitly but instead are accounted for non-parametrically. et al. & Parkhurst, J.O. The COVIDcast site from the Delphi group provides both R and Python APIs to access the SafeGraph Mobility Data. The approach does not require epidemiological parameters, such as the incubation period or \(R_0\), nor information on NPIs. Development of forecast models for COVID-19 hospital admissions using anonymized and aggregated mobile network data, Modelling the dynamic relationship between spread of infection and observed crowd movement patterns at large scale events, Human behaviour, NPI and mobility reduction effects on COVID-19 transmission in different countries of the world, Human mobility and infection from Covid-19 in the Osaka metropolitan area, Association of Republican partisanship with US citizens mobility during the first period of the COVID crisis, Tracking COVID-19 urban activity changes in the Middle East from nighttime lights, Mobile phone data reveal the effects of violence on internal displacement in Afghanistan, COVID-19 Open-Data a global-scale spatially granular meta-dataset for coronavirus disease, Combining graph neural networks and spatio-temporal disease models to improve the prediction of weekly COVID-19 cases in Germany, A proficient approach to forecast COVID-19 spread via optimized dynamic machine learning models, https://doi.org/10.1080/09669582.2020.1758708, https://www.oecd.org/coronavirus/en/#country-tracker, https://www.kaggle.com/jcyzag/covid19-lockdown-dates-by-country, https://docs.safegraph.com/docs/social-distancing-metrics, https://github.com/CSSEGISandData/COVID-19, http://creativecommons.org/licenses/by/4.0/. After showing that our model accurately fits case counts, we use it to study the equity and efficiency of fine-grained reopening strategies. Med.14 (2020). Wesolowski, A., Eagle, N., Noor, A. M., Snow, R. W. & Buckee, C. O. Both models are reduced-form models, commonly used in econometrics, that characterize the behavior of these variables without explicitly modeling the underlying mechanisms that link them (cf.2). It achieves this by capturing dynamics that are governed by many underlying processes that are unobserved by the modeler. All SafeGraph data is anonymized and aggregated. While this reduced-form approach does not provide the same epidemiological insight that more detailed models do, they demand less data and fewer assumptions. This is how we are able to model who is infected, where they are infected, and when they are infected. If there are multiple people visiting the same POI in the same hour, and some are infectious while others are susceptible, then our model predicts that there is some probability of new infections occurring. You signed in with another tab or window. We briefly summarize our methodology below. J. Assessing changes in commuting and individual mobility in major metropolitan areas in the united states during the covid-19 outbreak (2020). A chart showing SafeGraph's Shelter in Place Index score in Colorado during the course of the coronavirus pandemic. We do not specifically examine the impact of school reopenings because children under 13 are not well-tracked by our cell-phone mobility data, so we are not sure we can fully capture the risk of these places. Science 368, 638642 (2020). 2 and S1. Model with no mobility measures consistently over-predict the number of infections and drift away quickly from the observed data.