Influenza Outbreaks Forecasting in Russian Cities: Is Baroyan-rvachev Approach Still Applicable?

In the current work we assess the accuracy of the outbreak peak prediction expected in a fixed Russian city with the help of the populational SEIR models fitted to the past local outbreaks happened earlier in the country during the same season. This approach was successfully used in USSR by Baroyan and Rvachev to predict flu outbreaks throughout 1970's. However, in early 1980's it became increasingly inaccurate and due to that the approach was not applied since. Our aim is to find out, using the long-term data on acute respiratory infections incidence for Moscow, Saint Petersburg and Novosibirsk as an input, to what extent this method of outbreak peak prediction is applicable nowadays, and discuss the methods of increasing its accuracy. We found that SEIR population modeling is still valid for prediction of peak heights, but not peak days.