Effects of Physical Distancing Measures

UNICEF's big data initiative, Magic Box, studies and predicts the effects of mobility on Covid-19 responses

Kolo Magloire, a 10 years old boy and young reporter, with his 6 years old friend Chris Alex in Korhogo, in the North of Côte d'Ivoire.
UNICEF/UNI325661// Frank Dejongh

The COVID-19 pandemic poses an extraordinary challenge to the world, our societies, health care systems, and economies. Currently the virus has been confirmed in more than 200 countries and territories.  

In this context, many countries are using physical distancing policies (from school closures to travel restrictions or full lockdowns) as tools to reduce disease spread, looking to avoid (or flatten) the curves of cases and deaths, seen already in so many countries.  

While there is an obvious relation between reduced social contacts and the speed at which the disease spreads there is little knowledge and a general lack of tools to understand the secondary effects of the containment measures. For instance, how will social distancing measures effect our societies, what socioeconomic effects will they have, how suitable are they in vulnerable and impoverished contexts, and are they sustainable over time. 

Arkan, 9 (right), studies at home while keeping a safe distance from his sister Siwi (left) during the COVID-19 outbreak in Jakarta, Indonesia, on 29 March 2020. In March 2020, the Jakarta provincial government closed all schools in the city in an effort to curb the spread of the COVID-19 virus.
Arkan, 9 (right), studies at home while keeping a safe distance from his sister Siwi (left) during the COVID-19 outbreak in Jakarta, Indonesia. In March 2020, the Jakarta provincial government closed all schools in the city in an effort to curb the spread of the COVID-19 virus.

In this unprecedented situation there is a clear need for real time information. However, to leverage the full potential of Data Science, Big Data, Complex Systems Theory, Epidemic Modeling, and Computational Social Science requires joint efforts between scientific institutions, governments, and international organizations.

It is central to provide evidence and tools that allow for timely action, and for identifying the needs of the most vulnerable, in order to balance the severity of containment measures while mitigating the socioeconomic impacts that this pandemic will surely have. 

Magic Box

Through data and data science partnerships with private sector companies and leading research groups, Magic Box — UNICEF's big data initiative — is working to provide data, tools and insights that allow timely monitoring of physical distancing, evidence on the suitability and sustainability of mobility reductions for low income settings, and better models that allow a better understanding and balancing of the potential impact of these measures on the disease as well as on the underlying communities. 

We're currently producing insights for 10 UNICEF programme countries: Colombia, Cote d'Ivoire, India, Indonesia, Malaysia, Myanmar, Mozambique, Mexico, Nigeria, and Ukraine. We're working with partners all over the world to increase data and analytical capacity and forming collaborations with leading research groups and private sector companies to help fight this disease.  

We focus on: 

  1. Building tools to monitor and evaluate aggregated metrics of physical distance and human mobility, with special attention to the bias of big data 
  2. Developing methodologies to understand the suitability of containment measures (including mobility restrictions) for the most vulnerable contexts — low income settings, slums, refugee camps, etc.
  3. Disaggregating the impact of secondary effects across age, gender, and socioeconomic status  


Magic Box data 01 May
UNICEF Innovation

01 May 2020

This report includes insights for additional programme countries: Myanmar, Mozambique, and Ukraine. Moreover, the report includes an air pollution analysis for 10 programme countries, showing how air pollution has changed due to containment measures. The report also contains an extended exploration section including: the impact on work (USA, Spain), the sustainability of containment measures (Germany), and disaggregating mobility (into age and gender) in Tokyo, showing that young people are the ones that are most affected by lockdowns. 

Representativeness of human mobility data.
Representativeness of human mobility data


14 April 2020

Along with updates to the previous countries, this deck includes insights for Programme Countries Colombia and Cote d'Ivoire. It also contains an exploration section for countries currently experiencing an epidemic (USA) and countries ahead of the epidemic curve (Germany).

Measuring effects of containment measures, from human mobility data
Measuring effects of containment measures, from human mobility data


3 April 2020

Magic Box now includes analysis of data for 5 UNICEF Programme Countries: India, Indonesia, Malaysia, Mexico, and Nigeria; showing major change on movement behavior correlating with implemented policies. 

Estimation of epidemic peak.
Rachel Oidtman
Estimation of epidemic peak


30 March 2020

This report studies changes in movement across Germany and the USA, along with behavioral changes in the USA, to begin exploring how well physical distancing measures are working. It also uses human mobility patterns and disease characteristics toward epidemic forecasting for Spain and Brazil

Understanding COVID importation risk from global flight travels. Data from Amadeus.
Understanding COVID importation risk from global flight travels


03 Mar 2020

This reports explains its usage of global flight traffic in predicting the spread of coronavirus, and details how UNICEF has shifted efforts towards understanding the secondary impact of responses such as school and factory closures, and even panic, on societies. We outline doing proofs-of-concept to quantify these and discuss scaling up our data science capacity.

The MagicBox Science Team

Manuel Garcia-Herranz, Chief Scientist 

Dohyung Kim, Principal Researcher and Remote Sensing Lead 

Vedran Sekara, Principal Researcher and Machine Learning Lead 

Our Collaborators and Partners

Frank Schlosser 
Humboldt University Berlin & Robert Koch Institute 

Laura Alessandretti 
Section for Cognitive Systems, Technical University of Denmark  

Sune Lehmann 
DTU Compute, Technical University of Denmark  

Rachel Oidtman 
Perkins Lab, University of Notre Dame 

Alex Perkins 
Perkins Lab, University of Notre Dame 

Kelton Minor 
Center for Social Data Science, University of Copenhagen 

Esteban Moro 
Massachusetts Institute of Technology & Universidad Carlos III de Madrid 

Mohsen Bahrami  
Massachusetts Institute of Technology 

Piotr Sapiezynski 
Khoury College of Computer Sciences, Northeastern University 

Global Data Science Project for COVID-19 


Data and Funding Partner 


Data Partners




What We Need

The Mobility Data represented here is collected by means of different digital sources, including flight tickets, mobile phones and smartphones. Each of these datasets are biased in their own way; for instance not everybody has a smartphone — the most vulnerable, especially, are not well-represented in smartphone datasets.  

To make the analysis more representative, we need:

  • Data partnerships with local mobile network operators
  • Collaborations with local academics/groups to adapt the analysis to local contexts.