Fig 1.
Mobility and contact patterns in different place categories.
(A), Daily visitor counts to five place categories (restaurants & bars, retail, arts & entertainment, educational settings, and others) in NYC during 2020, as recorded in the foot traffic data. (B), Daily reported COVID-19 cases in NYC. (C), Geographical distribution of restaurants & bars across NYC neighborhoods (color). Stars and arrows highlight mobility links with over 1000 visitors per day from January 6, 2020 to March 1, 2020. Stars indicate self-links where residents visited restaurants & bars within their own neighborhoods. The maps were created using PYTHON using the shapefile publicly available at https://github.com/nychealth/coronavirus-data/tree/master/Geography-resources. This is a public repository by NYC Department of Health and Mental Hygiene. The term of use can be found here: https://github.com/nychealth/coronavirus-data. (D), Daily average crowdedness (daily visitor counts per square meter) at POIs for five place categories. (E), Daily average dwell time (minutes) at POIs for five place categories.
Fig 2.
Mobility patterns across 42 NYC neighborhoods in four place categories.
Daily average visitor counts (in log scale) from home neighborhoods (y-axis) to destination neighborhoods (x-axis) in restaurants & bars (A), retail (B), arts & entertainment (C), and educational settings (D). The five boroughs of NYC (the Bronx, Brooklyn, Manhattan, Queens, and Staten Island) are indicated on top of each heatmap. Foot traffic data from January 6, 2020 to March 1, 2020 were used, representing the period before the implementation of governmental interventions. To avoid numerical issues for , we visualized the quantity
, where
is the daily average visitor count.
Fig 3.
Model fitting to neighborhood-level COVID-19 case data.
(A), Simulations using model parameters estimated for the period from March 1, 2020 to December 13, 2020. Simulated cases were aggregated to the city level and are compared with the daily confirmed cases in NYC (red line). The shaded blue area represents the 95% credible interval, obtained from 500 independent simulations, without adding the Gaussian observation error (i.e., it only reflects the uncertainty in the latent deterministic trajectories). The solid lines represent the median of those 500 trajectories. (B), Simulations in four representative neighborhoods in the Bronx (upper left), Brooklyn (upper right), Manhattan (lower left), and Queens (lower right). Maps display the geographical locations of these neighborhoods. The maps were created using PYTHON using the shapefile publicly available at https://github.com/nychealth/coronavirus-data/tree/master/Geography-resources. This is a public repository by NYC Department of Health and Mental Hygiene. The term of use can be found here: https://github.com/nychealth/coronavirus-data.
Fig 4.
One-week ahead retrospective forecasts for neighborhood-level COVID-19 cases.
The behavior-driven forecasts are compared with three baseline models: (B1), a metapopulation model without place category-specific mobility but with seasonal forcing; (B2), the behavior-driven model without seasonal forcing; (B3), the behavior driven model with seasonal forcing but static mobility matrices, crowdedness, and dwell time. We present (A) the relative mean absolute percentage error (MAPE) (MAPEs of the behavior-driven model minus those of the baselines, with blue indicating better forecasts), (B) relative log score (log scores of the behavior-driven model minus those of the baselines, with blue indicating better forecasts), and (C) relative weighted interval score (WIS) (the ratio of the WIS scores of the behavior-driven model to those of the baselines, with blue indicating better forecasts) for all 42 neighborhoods from June 8, 2020 to December 13, 2020.