A custom collaboration with: Quintiles
Human health depends on a wide range of variables that are impossible to account for completely. But a simulation modeling a broad collection of data to examine how, or if, it interacts can enable us to see patterns at work that point toward the answers we seek. Recently launched at IBM’s Almaden Research Center, the SPLASH project is one such simulation.
“More and more, we were seeing complex interactions among systems where things didn’t go as expected, because we didn’t know how the components interacted,” says Paul Maglio, lead scientist on the SPLASH project. “So we wondered if we could understand the interactions by putting together models, simulations and data from the components that make up the larger systems.” Although the IBM scientists could have tried out this approach on a number of areas, they selected health.
“Here the complex system comes in because people have behaviors and attributes, environments have behaviors and attributes,” explains Maglio. Rather than look at so many variables across all health, however, Maglio and his colleagues needed some focus, and so they selected obesity. At first thought, the factors related to obesity might seem obvious: eat less and exercise more. “But it really isn’t that simple,” Maglio says. “Obesity depends on a complex set of interactions: the food system, a person’s economic status, food choices and modes of transportation.” So the challenge of SPLASH is to understand how these variables—and many others—interact. “Then we can know how to advise, say, a community on things to change to have the best odds of reducing obesity.”
To get to where the IBM project can advise a community, the researchers first need to build one. “We’re not experts in playgrounds or how people spend their leisure time or make food choices,” Maglio points out. “This requires a community effort.” In order to incorporate a wide variety of sources, Maglio and his colleagues are collaborating with experts in a range of fields who have pertinent data, models and simulations to contribute to the effort. In a year, the project leaders hope to build a preliminary model of obesity. “Obesity is just a starting point,” says Maglio. “Later, we might focus on another chronic disease or look outside the U.S.”
Besides a community of experts, this work will require an array of computer resources. “We don’t know yet whether this will take a mainframe, supercomputers, clusters or the cloud,” says Maglio. “We could need to integrate all of those.”