¹Lomonosov Moscow State University, Moscow, Russia
²Marchuk Institute of Numerical Mathematics, Moscow, Russia

Abstract. An incidence curve of acute respiratory infections  in Moscow has three picks between September and April and reaches its maximum in January- February. The emergence of  new strains of influenza A could account for only one pick a year. The most cases of common cold are caused by ubiquitous  low pathogenic viruses. In order to simulate weekly fluctuation of incidence rate of acute respiratory illnesses we developed an agent-based model. It contains 10 millions agents with such attributes as  sex, age, social status, levels of specific immune memory and lists of contacts. Each agent can contact with members of its household, colleagues or classmates. Through such contacts susceptible agent can be infected with one of seven circulating respiratory viruses.  Viruses differ in their immunologic properties and  assume to present influenza A virus, influenza B virus, parainfluenza,  adenovirus,  coronavirus, rhinovirus and respiratory syncytial virus. The rate of transmission depends on duration of contact, vulnerability of susceptible agent, infectivity of infected agent and air temperature. Proposed network of social interactions proved to be sufficiently detailed as it provided good fitting for observed incidence rate including periods of school holidays and winter public holidays. Additionally, the estimates of  basic reproductive rate for the viruses confirm that all these viruses except new strains of influenza A are relatively  harmless and unable to cause significant growth of acute respiratory infections morbidity.

Key words: agent-based model, epidemiology, respiratory infections, contact network.