Dirk Brockmann, an associate professor in the Robert R. McCormick School of Engineering and Applied Science, is changing the way we track the spread of human infectious diseases. He uses modern technology, such as social networking websites, to examine human mobility. His work, with colleague Lars Hufnagel, analyzing data collected from the website Wheresgeorge, which tracks the geographic circulation of individual dollar bills in the United States, was key in developing reliable models for pandemic disease forecast.
What is the focus of your research?
The broad picture is that we are focusing on understanding complex systems. One example is the spread of human infectious diseases. We’re using mathematical models to try to understand these phenomena better.
Related to this is research on human mobility. Nowadays there are all these devices that can track where we go, such as iPhones or GPS devices, and there are lots of social network websites that collect data. We can potentially understand patterns behind infectious diseases by using mathematical models or methods from theoretical physics to look at this data.
If you want to understand how H1N1 spreads in the U.S. there are basically two things that you need to know: how humans interact (because interactions lead to transmissions) and how people travel. If people are not traveling the disease cannot spread from A to B, and if people are not interacting the disease cannot proliferate in a population.
How is modern technology changing the study of infectious diseases?
Prior to the onset of pervasive data collection – data that is being collected on the Internet and by modern devices – we had to really guess about the underlying rules that govern our mobility patterns and interaction patterns. It was all based on intuition. Now we’re in this transition phase where we can collect all this data and measure how people interact and travel, and then use that to refine our models of disease dynamics.
Before everyone had a cell phone, we had the idea that if we know how people travel we can develop good models. But we didn’t know how people travel and we didn’t measure it. We had the intuitive notion that of course you travel short distances more frequently than long ones, but exactly how much?
We came across the Wheresgeorge website and thought that maybe we could see a trace of how humans travel by looking at how the money goes from A to B.
How did you implement these techniques when studying recent epidemics?
In 2004 we were asking ourselves, “Can we develop models that can predict the spread of a disease like SARS in the same way that you can make a weather prediction?” You can say tomorrow there’s going to be a 60 percent chance of rain and often that prediction works very well for a certain period of time. So the idea was to do something similar for emerging infectious diseases.
Then H1N1 came and we gathered data the first few days and started running these simulations and made predictions. For a few weeks it was very good and then it diverged, which was insightful. Just like a weather forecast, it works for a couple of weeks and then something else happens and you can no longer make a prediction.
The main reason was because people started reacting to the disease. What these models can do nowadays is only compute the worse case scenario, where no one does anything to contain the spread of the disease or people do not change their behavior. We have no way of incorporating this into a model yet.
That’s the big subject of the future – trying to understand how people react to the news of a disease.
What are you currently working on?
We’re trying to not only incorporate mobility behavior and social interactions, but also trying to understand how they change in response to such events. One way to understand this is to investigate how individuals change their communication patterns in response to large-scale events, like a bomb threat or earthquake. People start behaving differently on their social network sites and one can use that information as a model for how we change our interactions.
In many of these models, complex networks play a role. Many complex systems can be understood in terms of individual units or parts interacting in a network structure. Best example is a social network, where every individual is a node and a social tie is a link.
Imagine there is news spreading through some sort of social network. There are some people that have lots of connections – friends and peers. Then there are some that only have a few, and there are those who bridge one peer group with another.
You can imagine that news spreads in one group and then suddenly gets to the person that is functioning as a bridge and then it will get to the other group. Those people who have lots of connections are super spreaders in that they spread the news efficiently.
How do you apply this to human mobility?
If you visualize the worldwide air transportation network, the nodes are airports and they are connected by links. Some links are very strong, like New York and Chicago – they have a very strong link because lots of people travel between them.
If I run the simulation of an epidemic and I show this to you on a map you will see the hubs, such as Los Angeles, New York and Chicago, will get infected very quickly because lots of people live there and these places are very well connected.
Compare this to the spread of the Black Death in the 14th century, which started out in southern Europe and went like a wave over the entire continent. People didn’t travel long distances at that time, so it just went a step forward and infected the next villages and so on. It’s not like that any more.
Infectious diseases are not very old. The reason they could not be sustained in early human populations is because it wasn’t dense enough. Most of these viruses come from animals that are swarm animals, or exist in large populations, such as birds or cattle. For transmissible diseases you have to get two populations really close to each other in order to sustain it. So it was not a surprise that human infectious diseases emerged when large civilizations started to emerge. Ever since then the emergence rate of new disease has increased.
We now live in very dense situations and that facilitates the emergence and spread of new diseases. It’s an ongoing battle between the emergence of these diseases and us understanding better the way we interact and travel. So the better we understand this, the better we can contain a disease or mitigate the spread. It’s not like this problem is going to be solved very quickly. That’s the biggest challenge.
Learn more about Prof. Brockmann’s work here.