The Use of Mobility Data in Public Health Emergencies

Symbolic picture: the Use of Mobility Data in Public Health Emergencies

Published on: 11.11.2022
Author: Yara Abu Awad

This blog post is intended as a brief overview and has been adapted in part from the CrisisReady report by the same name. CrisisReady is a research response platform that collaborates with academic institutions, technology companies, and response agencies to embed data-driven decision making into local disaster planning. This platform was presented at a Data Science and AI for Public Good webinar organised by the Data Science Competence Center.

What is mobility data?

Mobility data is data that tracks human movement. It can come from any device connected to a cell phone network, the internet and/or GPS. The type of data ranges from Call Data Records (CDRs) to x-Data Records (xDRs – from mobile devices connected to the internet), vehicle GPS devices, geotagged social media data, bluetooth exchanges or data collected from smartphone apps.

What are the potential benefits?

The ideal uses of human mobility data in a public health emergency would be to: track the number and location of persons at risk, alert nearby hospitals about the volume of expected casualties and, help with epidemiological modeling. Unfortunately, there are some barriers to achieving this level of responsiveness. These include the challenges presented below.

What are the Challenges?

Access

Access to mobility data has to be negotiated from those who collect it. Namely: mobile network operators, companies that aggregate and broker data and smartphone apps (i.e. Meta can provide data collected via the Facebook app). These negotiations tend to take a long time, need clear regulation and may involve payment.

Precision

In some situations location data provided can be very imprecise. For example in a rural area where cell tower density is low, location may only be available within a wide radius. Data may also be temporally imprecise, offering infrequent location information. Furthermore, data providers may purposely provide less accurate location data in order to protect user privacy.

Representativeness

Owners of mobile devices who allow their data to be collected are typically not a representative or random sample of the population.

Technical expertise

The technical expertise to leverage mobility data may be inaccessible to most government agencies.

Trust & Privacy

Concerns over privacy and misuse (by malevolent actors) limit the release of mobility data. Misuse by government agencies (such as the use of mobility data by law enforcement agencies) will lead to an erosion of public trust and a lower willingness to share data.

Recommendations

The CrisisReady report provides action items (summarised below) in response to the above challenges.

Data Readiness

1. Introduce legislation to govern data use.
2. Launch a professional body to publish interoperability standards.
3. Generate context-specific consensus on aggregation and anonymisation.
4. Promote the development of standard contractual language for the use of mobility data.
5. Promote a cadre of data stewards within companies to guide the responsible use of the data for public good.
6. Develop, test, and disseminate use-cases on the application of differential privacy on human mobility data.
7. Include communities from whom the data are generated in defining the scope of use of data.

Methods Readiness

1. Develop frameworks for communicating bias and uncertainty.
2. Advance methods to address bias correction in human mobility data sets.
3. Support the development of standards to promote interoperability among the data sets.
4. Develop approaches to allow for the diverse anonymisation techniques used across technology companies.
5. Develop a framework for quantifying the potential for harm that acknowledges societal and political context.
6. Advocate for donors and national academies to invest resources, in preparation for and during emergencies, including allocating emergency funds.

Translational Readiness

1. Document and disseminate best practices targeting: 
   • technology companies, to incentivise them to keep sharing these data responsibly;
   • donors and national academies, to incentivise them to greatly invest resources;
   • policymakers, to socialise them to the potential use of these data products and to help response agencies guide purposeful, actionable analytic products;
   • general public, to improve transparency about the potential use and limitations of these data to promote informed adoption.
2. Fund and develop a cadre of "data bilinguals" embedded in public health and response agencies.
3. Fund and sustain a distributed network of researchers to support local regional collaborations of trained scientists and response agencies.

Conclusions

Human mobility data has great potential to enable a rapid and effective response to public health emergencies, so it is essential to overcome the challenges and barriers to its use.

Further information:

CrisisReady report: The Use of Human Mobility Data in Public Health Emergencies

Data Science and AI for Public Good webinar: CrisisReady

Last update 11.11.2022