Web Exclusives: Diseases
The Science of Simulating Disease Spread
Part 1: Why We Model Infectious Diseases
At the National Institute of General Medical Sciences, part of the National Institutes of Health, I manage a research project called the Models of Infectious Disease Agent Study, or MIDAS. I don't do the research myself, but I work with scientists at universities, companies and health agencies all over the world. Our job is to create models that help us understand the spread of infectious diseases and how to protect people.
Recent Infectious Diseases
Infectious diseases are caused by microbes or parasites that spread from one person to another. They affect a lot of people, some of whom will die. Several infectious diseases have been in the news; two that we are watching are:
- Staphlococcus aureus (Staph), a common bacterium found on people's skin and in the nose. While it usually doesn't create health problems, Staph sometimes causes diseases like impetigo, toxic shock syndrome or food poisoning. In the past decade, we've seen an increase in Staph infections that cannot be treated with usual antibiotics. The U.S. Centers for Disease Control and Prevention reports that in 1974, 2 percent of Staph infections were drug resistant. In 2004, that increased to 64 percent. In 2007, these drug-resistant microbes contributed to over 19,000 deaths in the United States.
- AIDS is a global disease caused by the human immunodeficiency virus (HIV). Over 33 million people in the world are infected with HIV and may go on to develop the disease AIDS. In 2007, over 2 million people died of HIV/AIDS.
Today, infectious diseases are causing more and more health problems. Just in the last few years, we've seen many new infectious outbreaks, including SARS, West Nile virus, hantavirus and now H1N1 influenza.
Infectious Disease Management
Managing an infectious disease starts with learning as much as you can about the disease and how it spreads. How does the virus or bacterium infect people—through the nose, in body fluids like saliva or blood or through a vector like mosquitoes or ticks? How soon after being infected do people get sick? How fast do they recover? What groups of people are most affected? What is the natural host of a disease-causing organism—birds, farm animals, mice? Does human behavior affect how the disease spreads? Does weather play a role?
Then you need to consider all the interventions you can use. Is there a vaccine? How effective is it? Will people be willing to be vaccinated? Are there drugs, like antivirals or antibiotics, that help? Are they readily available? Are they expensive? Are there behavioral or social interventions? For example, using bed nets can protect people from malaria-carrying mosquitoes, but will people use nets? Closing schools might protect people from influenza, but is it worth the cost? What about face masks or hand washing?
The question then becomes: How do you deploy resources to protect people from a specific disease?
This is where models come in.
Modeling Can Help
It would be impossible or unethical to do controlled experiments in human populations as a disease is emerging, but we can use mathematics, statistics and computational methods to think through the problems. In the computer, we can let a pandemic unfold many times under the same or different circumstances and see what happens.
A lot of times when we think about modeling, we think that models have to account for absolutely everything and give us the right answer. I've learned that this not how models work.
Scientists use models all the time, whether we are representing molecules with ball-and-stick kits, creating mathematical equations to study how cells work or using huge computers to predict earthquakes. Models help us understand how things work, now and in the future. Good models depend on good data, good tools and clear thinking. Just like a model airplane is not a real airplane, a mathematical or computational model is not reality. The model airplane leaves out details (like the passenger seats or engine) but keeps the important features (like wings so it can fly). Infectious disease models may leave out details (like the genome of the pathogen) but keep other important features (like how infectious a disease is).
Incompleteness doesn't mean modeling is useless. Modeling can really give you insights into how something works. For example, when I see a model of DNA, I can immediately understand how it copies itself. When people see models of how diseases spread, they often have one of those "Aha!" moments—"Oh, so that's how it works!" Modeling helps us conceptualize very complicated ideas, often by creating visuals, like movies that show how diseases spread throughout the country or even throughout the world.
Whether the models are "pencil and paper" or make use of the world's most powerful computers, they are helping us do excellent research to protect people from infectious diseases.
The MIDAS Program
MIDAS scientists build models using statistical tools, especially if they need to understand random or unpredictable events. They may use mathematical equations to explain how systems change over time. Often, modelers turn to large computers. MIDAS works closely with scientists in several supercomputing centers to build very detailed models of how disease spreads. These models incorporate information about the real population—such as the age, gender and health of people; where cities and towns are; and geographical features like mountains and rivers—to build what we call a synthetic population. We include information about where people work and go to school, what they do in their spare time, and whether they use public transportation. Once we have a synthetic population, we can add a disease, like pandemic H1N1 influenza, to see how it spreads. We then test interventions, like vaccination or closing schools, to see how well they might work. We can look at what would happen if people became afraid and stayed home. It's exactly like doing a laboratory experiment, except that it is all in the computer.
There are at least 50 scientists from different disciplines in MIDAS, and each one comes to the table with different expertise and perspective. They share their ideas and information and decide how to research questions. We can do much more as a group of experts than we can do individually.
We also critique one another's models and results. We don't want to miss anything important. We want to present the best information possible.
Read Part 2: Modeling the 2009 H1N1 Pandemic
Read Part 3: The Future of Infectious Disease Modeling
This article series by Irene Eckstrand first appeared on Year of Science 2009 as blog posts celebrating science and health.