In the face of a global influenza pandemic, the most stringent restrictions on travel are necessary to significantly reduce the cumulative number of people affected and to delay the peak of the outbreak. However, even low-efficacy vaccines, administered immediately from a stockpile, can significantly reduce attack rates. These and other findings were presented by a team of NCSA users, led by Neil Ferguson of Imperial College London, in April 2006 in Nature. As members of the National Institutes of Health's Models of Infectious Disease Agent Study (MIDAS), the team builds computer models of the spread of an influenza epidemic in human populations to estimate the effectiveness of different strategies to limit the spread of the disease. The models simulate every individual in the United States or other countries, together with every school and workplace and the journeys people make. They also simulate virus strains with different levels of transmissibility. Features of virus spread are based on historical data from previous influenza outbreaks. But, as the Nature paper points out, "Although... using data from past pandemics should be a priority, it will be impossible to predict the exact characteristics of any future pandemic virus...It will be imperative to collect the most detailed data on the clinical and epidemiological characteristics of a new virus and the impact of control measures early in the emergence of a pandemic and to analyze those data in real time to allow interventions to be tuned to match the virus the world faces." With this in mind, NCSA is helping members of MIDAS develop schemes for overflow computing, which allows new processors to be introduced into a calculation on the fly thus making on-demand computing of the sort described easier and more powerful. NCSA is also assisting in profiling the models' performance and optimizing the code being used. NCSA's Cobalt system is being used for these simulations because of the large shared memory available on that system. Findings from the Nature study include: - Household quarantine is the most effective "social distance measure," better than workplace- or school-based approaches. But it requires high compliance rates. - School closures have limited impact but can enhance other strategies. - Border restrictions can delay the peak of the pandemic by as many as six weeks, allowing time for more people to get vaccinated. - Domestic travel restrictions have almost no impact if they are followed by less than 75 percent of the population. Significant reductions in the spread of the virus require better than 90 percent compliance. These results are being used to assist policy makers at the highest levels of the U.S. and other governments. This research is supported by the National Institute of General Medical Sciences' MIDAS Program, the Medical Research Council, the Royal Society, and the Howard Hughes Medical Institute.