Modelling the spread of HIV

hiv-spread.jpg

Modelling the spread of HIV is a difficult business for many reasons: many people are unaware that they are infected, HIV can take a very long time to manifest itself within the body, and researchers are still unsure to what extent different population groups are involved in its transmission.

So it’s remarkable that Shan Mei at the University of Amsterdam and colleagues have been able to model the the HIV epidemic among the relatively small group of gay men in Amsterdam.

The trick they say is to master two approaches to modelling. First, create an agent-based model of human behavior and then create a realistic model of the network that links real people together in Amsterdam.

Mei and friends say they’ve done this for the gay population in Amsterdam and that their model shows a “good correspondence between the historical data of the Amsterdam cohort and the simulation results”.

They even say the model can predict the future trend of HIV prevalence in Amsterda. That’s a much bigger claim that needs to ne matched with evidence which is sadly lacking from this paper.

Mei and co demonstrate exactly what confidence they have in this claim by making no meaningful prediction whatsoever. Strange.

Ref: arxiv.org/abs/0812.1155 : Complex Agent Networks Explaining the HIV Epidemic Among Homosexual Men in Amsterdam

Comments are closed.