93163121
I wrote this once but blogger ate it. Maybe it�s a good thing as the first one was rather rambling.
I wrote this post over at DocBear�s SARS message board at Agonist.net
�I question the usefulness of any mathematical models for the spread of SARS. The only model that would seem to fit would be a network model. In network models, the most important parts are nodes that have many connections. To predict the behavior of the network you need to look at these nodes, not the whole network on general.
Superspreaders are clearly similar to these nodes, but we already know that the most important thing to do to control SARS is to stop superspreaders. Even the network model doesn�t give us very useful information. There are two many unknowns to make any useful predictions about doubling time. Will there be a large outbreak in the U.S., Africa, India, Russia? The actions of one superspreader could drastically change the future of the outbreak.
Phil B. asks a very good question, Is SARS being contained in Guangdong and if so how is it being controlled? The WHO gave Guangdong high praise for their efforts to control SARS, so I think it is possible that the doctors there might have done a good job. But then they might just be hiding a lot of cases. I’m also curious as to how VietNam was able to stop their outbreak. They seem to have done better than the Canadians.�
I got to thinking and I came up with my own network model of SARS spread.
The nodes in my network would be cites or centers of population. It would be best to have individuals be the nodes, but since I want a global simulation and the number of SARS patients is so low compared to the population, this would be impossible. In my network every node is connected and the strength of the connection is determined by the amount of travel between the two cities. Each node has a level of infection ranging from 0 not infected to 5 out of control epidemic. Each level has different properties. Each city also has a measure of the quality of healthcare.
The network changes by doing two calculations every turn (representing Days or weeks). First it calculates the level of infection in every city based on the cites previous infection and the quality of health care. The second calculation is done on all the connection, it calculates the probability of being infected by another node based on the other nodes level of infection and the strength of the connection.
The levels of infection are
0 no infection, can only be infected by another node, at which time the city will either go to level 1 or 2, 50/50 percent of the time, but also taking into account the cities quality of health care. Probability of infecting another city 0*
1 small contained infection (i.e. just a few people have it and they are all properly quarantined, like the U.S. now), probability of moving up/down/staying the same 3/17/80. infecting another city