Can we use computers to understand generosity? The behavior of crowds? The root of prejudice? Can we use computers to improve society? Researchers have recently made a lot of progress in developing computer models to improve our understanding of human behavior and the world we live in.
If you really think about it, each one of us is already a modeler. We create mental models every time we try to predict the outcome of a sporting event, an election or the stock market. But our mental models are based on limited facts and are skewed by our perspective and assumptions. Computer models, in contrast, rely on objective data. They can account for many more factors than the human brain, and they can be tested, adjusted and verified.
Computer models have many potential uses. Simulations of how people behave in the face of natural disasters could help prepare local and national officials to respond. Mathematically capturing the behavior of pests or beneficial insects could improve agricultural practices. And modeling social networks can help doctors develop better strategies to change unhealthy behaviors.
Using computer models, researchers are trying to predict—and propose ways to minimize—the effects of a future flu pandemic. Flu outbreaks in 1918, 1957 and 1968 killed millions worldwide. NIH-funded researchers looked at the measures that different health officials took in past pandemics and analyzed their effects. They used this information to build computer models and confirm the models could simulate what happened during past outbreaks. Then they used them to simulate an outbreak of pandemic flu as it spread throughout a large city.
The simulations helped researchers identify public responses that could significantly slow the spread of infection. Tactics like closing schools and giving anti-viral treatments, the models found, could give researchers more time to develop vaccines. The models are also helping researchers understand how people react to these public health measures and how to optimize their timing.
Computer models can help researchers combat other diseases, too. Last year, NIH-funded researchers created a computer model of cholera transmission in Matlab, Bangladesh. Cholera bacteria, which spread through contaminated water and food, can cause severe symptoms that include diarrhea, vomiting and leg cramps. The disease can lead to death by dehydration in a matter of hours if left untreated.
The computer simulation showed that cholera transmission could be controlled if about 50% of the population got an oral vaccine. Public health officials now know they could likely control cholera with a modest investment using a mass vaccination program. These types of models can help health officials figure out which vaccination strategies would work best in different settings.
Last year, NIH-funded researchers computerized weight, height and other data collected over a 32-year period from a socially intertwined network of over 12,000 adults. They found that friendships can have a crucial influence on a person’s weight. In fact, the likelihood of becoming obese increased by nearly 57% if a close friend had become obese.
This year, using the same technique, the scientists reported that close relationships exert a strong influence on smoking. The greatest effect was in married couples. When a husband or wife quit smoking, it reduced the chance of their spouse smoking by about 67%.
This research suggests that it may be possible to harness social networks to help people change behaviors, such as smoking, for the better.
No computer model is perfect. Even the best can only take into account the things that we know about and can measure. Modelers also routinely have to make decisions about what to include and exclude.
Researchers continue to design and test new computer models. As they improve, they give researchers new insights into the most effective ways to affect people’s health.