Simulation in Synthetic Populations
In 2003, Magnus Boman was promoted to full professor of Intelligent Software Services at KTH. The last few years, he has turned his attention more and more toward simulation and synthetic populations. Can the epidemic spread of a contagious disease be predicted? How could such a catastrophe be prevented? And if such an event should nonetheless occur, what would be the best methods for containing and dealing with it? These are some of the questions that synthetic populations used in simulation models help answer.
Magnus Boman divides his time between his two offices, at SICS (Swedish Institute of Computer Sciences) and KTH. When teaching, he can be found at the Forum building. He conducts his research, though ― both for KTH and SICS ― primarily at SICS, where everything is set up to promote research. His work revolves around creating models for predicting, preventing, and handling extensive and serious problems such as infectious diseases and catastrophes. For such research he develops models using synthetic populations.
Synthetic populations are not built from actual individuals but from information about groups/society obtained from various existing sources. The process of creating a “synthetic” individual involves borrowing profiles of real individuals and fashioning another profile that is sufficiently similar.
“With such synthetic populations we can study, for example, the spread of contagions in society. Because,” Boman points out, “we obviously can’t study this in reality by releasing a virus and observing how it spreads. Instead, the work requires a study environment, using a synthetic population generated from a real one.”
For some years, Boman has been working with Dr. Lisa Brouwers, a post-doctoral researcher at the Swedish Center for Infectious Disease Control. With support from the National Board of Health and Welfare, they have worked together on, for example, patterns in the spread of infectious diseases.
In 2007, Boman and Brouwers formed their own research group, SyntPop, to work with synthetic populations ― not limited to the area of epidemic research, but even for broader applications. There’s a keen interest in using synthetic populations in many areas ― medicine, mathematics, statistics, sociology, computer sciences; even the field of economics is exploring it.
“Here in Sweden we are far ahead of the pack in this area ― first in the in the world, actually,” tells Boman. “Available resources in Sweden provide good opportunities to create good synthetic populations. We have public statistics well known to be of a high quality.”
Boman formally worked as a post-doctoral fellow at the International Institute for Applied Systems Analysis (IIASA), an internationally recognized institute founded 35 years ago. Boman returns there occasionally to guest lecture.
“IIASA works with large human system issues, such as the climate, global warming, the spread of infectious diseases, HIV/AIDS, and world hunger,” reports Boman. “It’s an incredible place to do research. They have enormous political support for their work, and the OECD often advises the institute.” In the future, he anticipates a collaboration using synthetic populations.
“I’m hoping the EU will sponsor a project on synthetic populations,” says Boman. “Statistics Sweden [SCB, a state authority collecting statistics on the whole of Sweden] has such an amazing database of public statistics. In Sweden, we keep tabs on what’s going on in society. It’s a tremendous asset for use in, for example, pandemic research.”
A first project proposal was rejected because it lacked a European perspective. But Boman plans to resubmit an application and feels relatively confident of getting funding to conduct research at the European level.
Stimulating Cross-Disciplinary Work
Boman’s area of research is a relatively new field, which means there are no established experts to evaluate research project proposals. The field is multi-disciplinary and, quite frankly, there aren’t many multi-disciplinary experts out there. As a consequence, Boman develops a lot of his work in networks, working with other researchers within a variety of fields.
“I’ve written numerous articles in all sorts of contexts and together with a number of different researchers in assorted fields,” says Boman. “This could certainly work against me, but I think that most all of the work can be seen to fit under the umbrella of Synthetic Populations.”
Most recently, Boman turned his interest toward computer games. He’s looking at games as places for wide-scale simulations where different players meet one another ― that is, a kind of anonymous gathering. Boman thinks that through games it could be possible to learn how reality functions.
“I’m very interested in this area,” says Boman. “At a conference in Japan, I presented a paper, along with my colleague Stefan Johansson (Blekinge Tekniska Högskola), on using computer games instead of synthetic populations. The paper gained considerable media attention.”
Through a KTH Small Project grant, Boman began collaborating together with visiting professor Zary Segall ― for example, in the course “Think Wearable!” which focuses on product innovation. Boman looks forward to continuing his work with Segall if future opportunities allow.
“Watching him teach has been very inspirational,” says Boman. “The pedagogical approach he brings to his classroom should be included in all courses.”
Boman cites his work on several assignments with the American company SunGard as one of the more stimulating projects his been involved in. Among other things, SunGard, with a staff of around 200 in its Swedish office in the World Trade Center in Stockholm, provides agents for financial transactions.
“It’s a progressive company that is out to make money. At the same time, though,” Boman, notes, “the people there recognize that research that improves knowledge about AI, machine learning, agents, etc., is vital to financial success.”
Unlike many other researchers, Boman seems confident about financial support for his work. “Prospects seem quite good for the future for my research. I’m writing articles and proposals, and things are moving along smoothly for the time being,” concludes a discernibly satisfied Boman.
Family: Wife, two children, and two cats
SyntPop (= synthetic populations)
SyntPop is an umbrella organization for research, seminars, and courses. Heading it are two senior researchers, Magnus Boman and Lisa Brouwers. The group also includes four doctoral students plus former doctoral students. Clients include SICS, the National Board of Health and Welfare, and the Swedish Institute for Infectious Disease Control.
Professor Magnus Boman and Dr. Lisa Brouwers founded the organization in 2007 to meet the demand for investigations into computer-generated populations of individuals. Of particular interest are studies using such populations in simulations of complex processes, such as epidemic spread in space and time. While the concept of “individuals” normally pertains to people, it could be generalized to denote animals, robots, viruses, or even trees. But processes governing spread are the group’s primary concern.
In general, modeling people and their actions is very difficult, and every study needs to be carefully scrutinized. When using data on real people, such as demographics, issues of anonymity and personal integrity for each individual reign supreme. Even the handling of such databases requires extreme care ― and often special permits ― for running simulations, because of the ethical dimension. If instead a synthetic population can be generated with the same, or at least very similar, properties as the original database, the synthetic population can replace the original one in simulations and other forms of manipulation.
The International Institute for Applied Systems Analysis (IIASA; iiasa.ac.at)
The Swedish Research Council Formas, the Swedish organ for IIASA cooperation (www.formas.se)