This article demonstrates successful application of several syndromic surveillance features and approaches: * Use of a nontraditional data source (telehealth calls) for annual influenza surveillance * Use of residuals of a reasonably simple data model to remove systematic behavior * Epidemiology-driven, age-specific threshold values supported by application to historic data The authors present evidence that these methods can yield significant timeliness advantage in influenza surveillance, The discussion is clear and objective, and the study data are sufficiently explained.

Previous authors have modeled syndromic data with autoregressive time series analysis and applied control charts to the residuals going back to Williamson and Van Brackle in the 1990's, but the SARIMA model used by these authors goes farther and yields good detection performance on data from several cities. The use of total visits as a fixed-effects covariate along with the SARIMA error model seems to be an effective combination for time series both with and without substantial day-of-week effects. (Other authors monitor the quotient of syndromic counts and total visits as surrogate rates, and these 2 uses of the total visits should be compared.) These authors use some terminology that is nonstandard in the surveillance community but not difficult. Their residual filters may be understood as control chart approaches. The simulation methods are plausible but not satisfying from a practical viewpoint. The statistical discussion is detailed, novel, and contains helpful insights for other modelers.

This article reflects on a "virtual" epidemic that hit the massively multi-player online game (MMORPG), "World of Warcraft" (WOW). The disease, "Corrupted Blood" inadvertently spread rapidly throughout the game attacking players that it was never intended for and infecting animal as well as "human" characters in the game. This opinion piece reflects on how the use of "real" humans in a MMORPG might be superior to the use of computer agents in a simulation. While I agree that it is interesting to observe MMORPGs and the spread of "virtual" diseases within it, I found that while the article spent a great deal of time describing the current state of these games such as WOW, there was not a rigorous analysis of the true situation that the game is simulating. The authors make statements ascribing realistic player reactions to infectious disease situations in the game without any real evidence that this is the case. The articles about MMORPGs referenced are not in-depth studies of the psychology of these games, nor did they analyze the requirements of a game that is a good simulation of disease spread. Agreed, the use of game technology has been neglected in some simulation work but there are many aspects to simulation that are not requirements for good on-line games and this whole issue was glossed over in the paper. The level of understanding of game playing psychology and sociology was also limited. If I am playing a dwarf or a warrior in WOW will I really react to disease in the same way that I would in real life? The authors do admit that people play these games for entertainment and escapism but there was no attempt to truly analyze how this would impact on findings from these game situations. The basic premise of this opinion piece is interesting - what can disease spread simulations research learn from on-line game communities and can this be applied to produce a "better" simulator but after that the article provides mostly just a description of WOW and its Corrupted Blood epidemic with very few concrete ideas about how this could be adapted for real disease spread study. I was intrigued that they mentioned the fact that the disease in WOW spread rapidly because players used the "Teleport" feature to travel long distances. But they failed to make the connection that it might not just be the ability to travel long distances but the fact that players probably teleported to major centres of activity and thus this reminds me of the work in Scale-Free Networks that has demonstrated mathematically that travel or contact networks that contain "hubs" (nodes that have many connections) are excellent conduits for the rapid spread of disease. There are some excellent papers that examine this in the world of computer viruses and the world of human STDs. In conclusion, this is an interesting opinion piece but I would love to see someone take these ideas and write a more rigorous examination of the possible use of human agents in simulations of disease spread.

The authors model the spread of SARS through nosocomial transmission using a stochastic model with parameters based on previously reported Gamma distributions. The model can then be used to analyze two important relationships: (1) the effect of the non-isolation period on incident and prevalent periods and (2) the effect of the spreading ratio outside the hospital on the number of incident and prevalent periods.

Novel technique for outbreak detection leveraging relationships between data sources. this allows for handling situations such as population surges (e.g.olympics) and public health crises (e.g.: worried well during avian flu. "This article presents a method for combining multiple data streams by monitoring quotients of all pairs of streams in a surveillance network. The results compare these networks to more traditional monitoring methods applied to realistic data effects of simulated outbreaks. The method as given requires years of historic data and is applied only to outbreaks at a single facility, but these are not inherent limitations."

Review of research on the use of syndromic data for influenza surveillance

"This article seeks a statistical method for epidemic monitoring and to 'confirm that the peak has occurred and the decline has started' in a manner that is robust to year-to-year variations. A concept using peak monotonicity properties is compared to more traditional methods."