Big Data in Crises: Predicting the Future

Little ‘Data’, aNto on Flickr CC by A

Gathering, processing and interpreting data sets is what runs the modern global economy. Everything from your weekly online supermarket grocery shop, to how the shelves get stacked with goods delivered by cargo ships from all across the world. Teams of statisticians make their daily bread from finding more and more sophisticated ways of predicting human behaviour.

Complex mathematical modelling is also being used in humanitarian emergency situations to prevent further loss of life. I’m not just referring to how aid is delivered through more efficient supply chains, but to how the mapping of crises and outbreaks of diseases is used to instigate the correct response and predict how the spread is evolving.

Dr Sebastian Funk at the London School of Hygiene and Tropical medicine is a leader in the field and was at the centre of important work to map the Ebola outbreak in West Africa. By processing data collected from the affected area Dr Funk and peers across the world were able to look almost 6 weeks into the future, with 95% confidence of the first week.

Finding a ‘magic bullet’ or key to suppressing an outbreak is time sensitive – one must collect enough quality data to make sure that the models can be accurate, but when people’s lives are at stake conclusions need be drawn quickly.

It was found that ‘most people infected with the deadly virus became ill through contact with a small number of so-called ‘superspreaders’ and ‘if superspreading had been under control, about two-thirds of Ebola cases could have been avoided’.

Here he is speaking to RFI’s English service

Dr Funk’s work was reliant on effective feedback on the ground. He knew that whilst cases might be dropping off, there are unreported areas and if the superspreaders had been identified earlier, lives could have been saved.

The unprecedented rise of smartphone and social media has changed the data landscape. Agencies can have access to crowd-sourced information, which taken together can be highly accurate.

This was already happening during the Ebola outbreak – ‘When epidemiological data are scarce, social media and Internet reports can be reliable tools for forecasting infectious disease outbreaks’, from findings published in the Journal of Infectious Diseases.

Patrick Meier advocates the use of geographically mapping social media posts –  in Digital Humanitarians: how BIG DATA is changing the face of humanitarian response he describes crisis mapping of the 2010 Haitian earthquake and Libyan political turmoil in 2011 known as the Arab Spring. This approach allows aid workers to have a more complete live overview of the situation that’s constantly updated.

Libya Crisis Map Deployment 2011 Report

The crowd-sourced reports are collated by a volunteer team that work 24hrs a day, corroborating information from social media sources. It led to a significant change in the response, something which was praised by UNOCHA, if not without some reservations.

In this case, the role of the human as an arbiter of trustworthiness remains a significant undertaking. Even with pleas like: ‘we just need to make sure you’re not Gaddafi!…we are not Facebook!’ (Meier, p.125) for people to declare their background information, the digital gathering of sources, even on a big scale, still has to have an element of journalistic rigour.

The future will be to use artificial intelligence to perform checks that people simply don’t have the capacity to do, given the volume of information coming in. Systems are being developed that analyse qualitative rather than quantitative qualities of posts, allowing computers to detect false rumours and unrelated background noise.

With accurate models, the prediction capabilities in future outbreaks of disease or disaster will certainly be enhanced, leading to an untold impact on lives saved.


Leave your comments or tweet @data_bigly if you want to join the conversation.

References. Links in text plus:

Mier, Patrick. Digital Humanitarians: How BIG DATA Is Changing the Face of Humanitarian Response CRC Press, London. 2015

The Women’s March – Local Going Global

Women’s March in Washington DC, January 21, 2017. © Hanna Rhodin

The body of people forced itself forward, towards the White House. The air was crisp and the atmosphere was vibrant, optimistic, to the point where it almost felt like you could touch it. Pink hats were worn proudly and signs were raised up high. Saying “I’m with her” surrounded by arrows, a picture of a cat saying “Grabs back”, the creativity and the messages were many, touching upon gender issues, corruption, disbelief in the new president, immigration, refugees, and of hope for a brighter future. A woman was holding a big blue flag with yellow stars on it. When asked what the flag represented she said “Alaska”. All around people struck conversations with strangers from all over the United States, all gathered in the nation’s capital, Washington DC, on January 21, 2017 to have their voice heard.

I first heard about the Women’s March on Facebook. I clicked that I was “Interested” in the event. Weeks later my immigrant friends and I walked along the streets in Washington DC toward the march. Social Media was how we all found out about the event, and soon thereafter the march got traction in the mainstream media. It seems that no longer is social media something we can discredit from affecting people’s actions and opinions, not to mention politically. Aday, Farrel and Lynch et al. writes “New media, such as blogs, Twitter, Facebook, and YouTube, have played a major role in episodes of contentious political action. They are often described as important tools for activists seeking to replace authoritarian regimes and to promote freedom and democracy, and they have been lauded for their democratizing potential.” (Aday, Farrel, & Lynch et al. 2010:3). Over 200 000 people clicked “Going” to The Women’s March in DC, over 200 000 clicked that they were interested in the event, and in reality, estimates show 470,000 to 680,000 participants.

Jeremy Pressman, a professor of political science at the University of Connecticut, and Erica Chenoweth, a professor at the University of Denver and an expert on nonviolent protest, collaborated and created a spreadsheet open to the public. They gathered data from coverage and news of marches around the world. Whilst their best guess is a total of 4,157,898, their low estimate versus high estimate ranges from 3,267,134 to 5,246,670. To eventually settle the question, artificial intelligence may come to the rescue providing advanced technology to crowd counting, as organizers often have a reason to exaggerate in order to convey an even more impressive turnout.

Can social media take all the credit for creating the turnout for the Women’s March? Not necessarily. In 1995, before internet had made it is breakthrough in daily life the Million Man March in Washington DC, attracted an estimated 400,000 to 600,000 participants. It is possible that the Women’s March on its own would still gather a large support with our without new media, but that it is new media alone, we cannot take for granted as more factors likely would play a part. However, social media can be a very powerful and important tool, it can also lower the communicational transaction cost (Aday, Farrell & Lynch, 2010:10f).

Regardless of the actual turnout, the Women’s March in Washington DC, in many other cities and towns around the United States, and all over the world, was a powerful statement in unity and in number, and a testament to new media being used to mobilize, organize, and democratize.

References

Aday, S., Farrell, H., Lynch, M. et al. 2010: Blogs and Bullets: New Media in Contentious Politics, Washington, DC: United States Institute of Peace.

Captain, Sean. January 20, 2017. ‘The Science and Politics of Counting The Inauguration and Women’s March’. Fast Company. Retrieved February 25, 2017. https://www.fastcompany.com/3067376/fast-cities/the-science-and-politics-of-counting-the-crowds-at-the-inauguration-and-womens-m

Janofsky, Michael. October 21, 1995. ‘Federal Parks Chief Calls ‘Million Man’ Count Low’. The New York Times. Retrieved February 25, 2017. http://www.nytimes.com/1995/10/21/us/federal-parks-chief-calls-million-man-count-low.html

Pressman, Jeremy, and Chenoweth, Erica. 2017. Crowd Estimates, 1.21.2017. Retrieved February 25, 2017. https://docs.google.com/spreadsheets/d/1xa0iLqYKz8x9Yc_rfhtmSOJQ2EGgeUVjvV4A8LsIaxY/htmlview?sle=true#gid=0  

The Women’s March. Facebook Page. Retrieved February 25, 2017. https://www.facebook.com/events/2169332969958991/?active_tab=discussion

Featuring WPMU Bloglist Widget by YD WordPress Developer