[Paper Review] Mining Google and Apple mobility data: Twenty-one shades of European social distancing measures for COVID-19

Giacomo Cacciapaglia, C. C. Osakwe C.+1 more|arXiv (Cornell University)|Aug 5, 2020

COVID-19 epidemiological studies5 citations

TL;DR

This study uses Google and Apple mobility data to analyze social distancing effects across 21 European countries during the first COVID-19 wave. It identifies a consistent two to five-week delay in infection rate decline following mobility reductions, enabling classification of national distancing intensity and linking it to pandemic dynamics.

ABSTRACT

We employ the mobility data released by Google and Apple to investigate the effects of social distancing on the spreading dynamics of COVID-19 in Europe. We identify and quantify different degrees of social distancing and characterise their imprint on the first wave of the pandemic. The analysis allows us to classify countries according to their level of mobility. Furthermore we identify a negative change in the infection rate occurring two to five weeks after the onset of mobility reduction for the European countries studied here.

Motivation & Objective

• To quantify the impact of varying social distancing measures on COVID-19 transmission dynamics across Europe.
• To classify European countries based on their mobility reduction intensity using anonymized smartphone data.
• To investigate the temporal relationship between mobility changes and subsequent changes in infection rates.
• To identify a lagged effect of social distancing on epidemic trends, particularly in the first pandemic wave.

Proposed method

• Utilizes anonymized, aggregated mobility data from Google's Community Mobility Reports and Apple's Mobility Trends Reports.
• Measures changes in mobility relative to a pre-pandemic baseline for each country.
• Applies time-series analysis to correlate mobility trends with reported infection rates across 21 European countries.
• Identifies the lag between mobility reduction and observed changes in infection rate using cross-correlation or similar temporal analysis.
• Classifies countries into distinct levels of social distancing based on the magnitude and duration of mobility reduction.

Experimental results

Research questions

• RQ1 How do different levels of mobility reduction correlate with changes in COVID-19 infection rates across European countries?
• RQ2 What is the time lag between the onset of social distancing and the observed decline in infection rates?
• RQ3 Can mobility data from Google and Apple reliably classify national responses to the pandemic in terms of social distancing intensity?
• RQ4 Do countries with more aggressive mobility reductions exhibit a more significant and timely decline in infection rates?

Key findings

• A negative change in the infection rate consistently occurred two to five weeks after the onset of mobility reduction across European countries.
• Countries could be systematically classified into distinct levels of social distancing based on the magnitude of mobility reduction.
• The study identifies a clear temporal delay between mobility changes and changes in infection trends, supporting the effectiveness of delayed public health interventions.
• The analysis reveals that mobility data from Google and Apple provides a reliable proxy for measuring population-level social distancing behavior.
• The observed lag suggests that the impact of social distancing policies on transmission dynamics is not immediate but becomes evident within a few weeks.