Can Big Data Put the Brakes on Ebola?
Big data analytics have never before been used by humanitarian workers operating out of disaster zones. With the U.S.-based Centers for Disease Control estimating that 1.4 million new infections from the current Ebola outbreak will occur, many are asking if Big Data can curb the epidemic. In this article, we profile 5 contributions Big Data is making in the current fight against the deadly virus, and look forward to see what the aftermath will look like.
Researchers used to rely on techniques like counting heads at bus stations and asking sick people where they’d been traveling. HealthMap is an open-source data project, essentially a search engine for tracking diseases and identifying symptoms. Tracking not just Ebola but other disease trends – from malaria outbreaks to tick bites – Healthmap shows the most recent cases and locations where a disease occurs, along with all the conversations taking place about those diseases in real time. Its algorithm, designed by John Brownstein, professor of pediatrics at Harvard Medical School and Clark Freifeld, software developer at the Children’s Hospital Informatics Program, scours the Internet for specific keywords like “fever.” It sorts them by topic, excluding irrelevant phrases like “Bieber fever”. Real-time Ebola maps like this have real potential for speeding diagnosis. “If a patient has a travel history, this can help a clinician inform a diagnosis,” explains Brownstein.
2. Contact and Case Tracking
To get a handle on case-tracking – or, crucially, contact-tracking – the CDC has been using the Epi Info viral hemorrhagic fever (VHF) application, an open-source tool to speed up the process of finding people exposed to the deadly virus. EPI VHF gathers epidemiologic, clinical and laboratory data on every case. It was designed for outbreaks of viral hemorrhagic fevers: Ebola, Marburg, Rift Valley, Lassa, and Crimean-Congo. Field workers can write tailored questionnaires, and the app will produce detailed databases, maps and graphs according to the answers they receive. The app also works in areas with limited access to the web, since once it’s downloaded it doesn’t require an internet connection.
3. Big Data in Airports
Many airports are equipped with thermal imaging scanners that can detect a person with a body temperature higher than 38.6° Celsius, or 101.5° Fahrenheit, a potential sign of Ebola. While it might not be legal to stop or arrest such a person, this globally-collected data can show us where sick people are flying. Any unusual patterns could be a tip-off to an outbreak of an Ebola-related illness.
The technology was put to use in Hong Kong, South Korea and Singapore in the 2003 SARS epidemic and later during the outbreak of H1N1 swine flu in 2009. The advantage is that thermal imaging scanners are already present in airports across the world.
Cellphones – ubiquitous even in poor countries – can play a key role. All cell phones “ping” nearby towers with a unique ID number to announce their presence. In this way, mobile carriers amass huge databases containing fine-grained information on population movements and social patterns. In Senegal, Orange Telecom released anonymized, aggregated data from 150,000 phones to Flowminder , a Swedish nonprofit.
Flowminder built models of population movements using this and an earlier data set from Ivory Coast, also released by Orange. Unusual movement patterns among the cellphone users could signal an outbreak and predict where the disease is going to strike next.
Linus Bengtsson, a medical doctor and Flowminder’s co-founder, cautions that the model is essentially a first draft, and that it doesn’t include real-time data. But “In countries that already have epidemics,” he says, “this is the best estimate we can do of what mobility will look like. This can give the sense of the radius people tend to travel around.”
5. Predictive Computer Simulations
Computer simulations, similar to weather forecasting, can help predict the path of Ebola. Northeastern University Professor of Physics, Alessandro Vespignani, is working on Ebola modeling, predicting how air traffic affects the spread of the disease. His team at the Laboratory for the Modeling of Biological and Socio-technical Systems uses a high-resolution map of human populations (3,300 locations in 220 countries) and adds daily airline traffic, plus disease dynamics, such as incubation time and the fact that only symptomatic patients transmit the disease. “Since we have this model,” Vespignani says, “we can assess the probability of getting an infectious individual in countries across the world.” As conditions change, the model is revised and simulations are re-run. To make accurate predictions, it needs to be regularly updated with the number of cases and deaths at each geographical area.
Just as big data is making an impact on the handling of the Ebola crisis, it seems Ebola will change the way big data is used. The analytical challenges remain considerable. Data is collected in the field in often chaotic conditions, from inaccessible villages, amidst weak or non-existent health systems. Big data sourced from the Internet has the potential to be plagued by deceptive information and the difficulties of understanding that data in languages other than English. Privacy issues must be resolved, such as when using mobile phone data.
Ebola means big data is no longer just the realm of the business sector and much is to be learned. Humanitarian and public health organizations are recognizing its true potential. Predictive analytics enables vast swathes of information to be aggregated and filtered whilst removing irrelevant information along the way. Standard analysis and reporting cannot do this.
Predictive analytic solutions will definitely play a part in whatever comes next, after Ebola, according to the experts.
“Big data can certainly help in disease monitoring and sentinel,” says Yiliang Zhu, professor of Epidemiology and Biostatistics at the University of South Florida. What those managing the Ebola crisis urgently need is the powerful, sophisticated analytics that currently allow private companies to zoom in on winning business decisions. The challenge for those combating the Ebola epidemic is to urgently develop more predictive, forecasting analytics to save time, and ultimately to save lives.