Como foi possível errar tanto ao fazer predições para a gripe suína?

Vespignani é um bom físico estatístico. Então realmente não entendi o que está ocorrendo.

Número de hoje (10 de maio, ou seja, 13 dias depois da "worst case scenario") no site do CDC: 2532 casos, 44 estados afetados e 3 mortes. Isso é um "best case scenario", porque vários estados pararam de fazer testes. Um simples fit exponencial fornece 10.000 casos para 14 de maio...

Swine Flu: Statistical Model Predicts 1,000 Cases In U.S. Within Three Weeks

ScienceDaily (Apr. 29, 2009) — Indiana University Rudy Professor of Informatics Alessandro Vespignani, an internationally recognized expert on the statistical analysis and computer modeling of epidemics, said two different swine influenza infection models generated on April 27 both predict about 1,000 cases in the United States within three weeks.

There had been 40 cases of swine influenza (H1N1) reported in the U.S. as of 6 a.m., April 28, according to the national Centers for Disease Control and Prevention.

"But this is a worst-case scenario, as we are always working in a worst-case scenario setting," he said of the models performed at IU and Northwestern University. "What we are finding is that this is not a panic situation and that this thing is not ramping up in some crazy way. Right now we are confident that in the next few days things will be more optimistic."

That optimism is due in part to actions taken worldwide, such as the medical alert in Mexico, school closures in Texas, World Health Organization warnings, increased controls at international airports and the availability of an anti-viral drug for treatment. The next 72 hours will be critical, he predicted, and models could change as often as every 12 to 24 hours, based on worldwide events.

Government May Be Slow To React

David Orentlicher, M.D. and J.D., is co-director of the Center for Law and Health at the IU School of Law-Indianapolis. He said governments can be slow to react to threatened pandemics for two reasons. First, public health departments and programs tend to be seriously underfunded, and that can make it difficult to detect public health threats early or to mobilize responses to limit the spread of the threats quickly. Second, effective public health strategies can disrupt economic activity (e.g., if travel is restricted), and governments can be reluctant to implement needed public health measures when doing so will have undesirable economic effects.

Orentlicher also noted that economic considerations may drive the public to compromise the public health response. "People may find it difficult to stay home and forego wages when they may be sick or even when they are subject to quarantine," said Orentlicher. "It's important therefore to have provisions for job protection. It also can be important to have government programs to maintain people's income if they must stay home for an extended period and they don't have sick days or vacation time to fall back on."

PS: Foto com oscilações sazonais que poderiam ser o equivalente das oscilações observadas em redes de elementos excitáveis (Kinouchi and Copelli, Nature Physics 2006).