Using data to enhance collaborative crisis management

What are the key challenges in crisis management technology that can be better tackled by adopting an ethical approach?

Challenges in using data for crisis response

Crises and disasters are increasingly understood to be complex events that require data from across borders and disciplinary understandings to build complete and actionable situational awareness. To support the multifaceted data needs to build situational awareness, information technology (IT) are often sought as solutions. These solutions are often based in machine learning and data analytics to help filter and gain useful insights from the big, diverse, and inconsistent data gathered through these systems.

IT platforms can help collect the different types and sources of data in one place, from tools like satellite and drone data, CCTV, alarm sensors in buildings, GPS worn by responders at the scenes, or numbers and locations of ambulances.

Put together, these data can help make a situational understanding of risk that can inform a “common” approach and make working together across a range of boundaries more feasible.

That said, it is well acknowledged among disaster practitioners that how they approach risk is not the same on two sides of a border or even one regional agency to the next. Different risk assessment practices have different data needs and place different priorities upon what is considered relevant and accurate. Mixing these together into common, bucket, categories can lead to misrepresentations, unintentional bias, function creep, increased fragmentation, and new forms of liabilities and power struggles for those using the IT. Moreover, the analytics that drive the IT is often quite opaque for anyone without years of training, with plenty of news reports showing how even the best-intentioned analytics have resulted in racial bias and gender prejudice.

This is often why willingness to share does not lead to actual collaboration.

These issues need careful consideration and to be addressed head-on. This is why Trilateral is currently working with disaster planners and technology designers to develop both IT and organisational tools to help address these ethical implications in the IN-PREP project.

IN-PREP project response

IN-PREP is an EC funded project to design an Integrated next-generation PREParedness programme for improving effective inter-organisational response capacity in complex environments of disasters and causes of crises.

IN-PREP specifically creates

  • A common technological and data structure to connect a large quantity of diverse data
  • Data analytics and machine learning framework to help process that big data and provide insights relevant to all parties
  • Use it to make realistic scenarios for cross-border training and planning.

The project is made up of 20 partners representing 7 EU member states, from academia, industry, emergency management, SMEs. This has provided a good base for bringing together conflicting understandings of risk assessment, different techniques for disaster planning, and varied data practices.

To support the designers in identifying and taking these issues into account, the project incorporates an ethical and privacy impact assessment, led by Trilateral.

IN-PREP co-design approach: designing new technology with end-users

Generally speaking, there is an understanding that personal data should be treated carefully. However, bringing designers and end-users together to actively brainstorm other ethical issues where practice and technology meet has raised less visible but equally important challenges. These included, among others, implicit bias in local data structures, the ethics of responsibility for data, and power struggles around data access.

The more end-users have learned about how the analytics would work, they realised they cannot be reasonably expected to have the tools or capacities to notice when implicit bias – based deeply in cultural or societal understandings, challenges, and current affairs — is used to filter information to manage informational overload or fill in missing data.

It also became clear that there were concerns about responsibility. Who is responsible for providing the data? Maintaining and updating it? Who initiates the sharing? Getting it to the right person? Interpreting it correctly? Ensuring accuracy?

By working with end-users, designers have been able to gain a new awareness of how disaster planners and responders need to interact with the data even before using a tool, and are now aiming to build clear definitions of responsibility within the use within IN-PREP solutions.

At an even more basic level, through the impact assessment, it has become clear that end-users are concerned about not just their preparedness plans for action, but what these systems mean for data preparedness.

The amount of resources — funding, training, personnel, time – necessary to implement the background work so that data can be shared through such a system is daunting.

New political power struggles can emerge between those, for example, who can get data freely from government satellites and offices and those who have to build relationships with private companies to procure data.

An important take-away

Our work has made us acutely aware that data practitioners need to avoid creating new divides between the haves and have nots and avoid discriminating against those without access to such resources. Data can enhance crisis response and the design work on new technology can balance what is possible with innovative data analytics and infrastructures with what is possible practically in terms of disaster planner time, knowledge, and skills, European-wide.

For more information on this research area, please contact our team:

Katrina Petersen, Senior Research Analyst at Trilateral Research

Katrina Petersen TRI


Risk Assessment Methodology Report

you can view the Executive Summary and Table of contents of the Risk Assessment Methodology Report.

Please sign up to the Solebay mailing list to download the Full Solebay project report.