Predicting the Influenza Season 2013-2014 in Italy

Background Influenza-like illness in Europe is monitored by means of nationwide networks of sentinel General Practitioners (GPs) detecting only medically attended ILI cases. Internet-based community surveillance of influenza is an increasingly-used tool in epidemiological data collection, especially for recruitment and follow-up of large cohorts as it can help to determine the disease burden, time trends and seasonality, and to characterise care-seeking and treatment behaviour as well as patterns of absenteeism. Internet-based systems can thus enhance traditional GP-based surveillance and support the interpretation of recorded data, both for pandemic and seasonal Objectives An Internet-based community surveillance of influenza-like-illness (ILI) can become a means of gathering epidemiological data from the general population in a participatory fashion. In this work we show how the high geographical and temporal resolution of the web-based participatory surveillance of influenza allows the feed of predictive models for the weekly forecast of influenza. Methods A network of Internet-based surveillance systems, called Influenzanet (www.influenzanet.eu), has been active in ten European countries (the Netherlands, Belgium, Portugal, Italy, the United Kingdom, Sweden, France and Spain, joined by Denmark and Ireland since 2013-2014), each using the same platform with the aim of gathering epidemiological data across different countries in a standardized way and with the participation of a self-selected cohort of volunteers followed over the influenza season. Upon registration, users are invited to fill in a background questionnaire containing various socio-demographic, medical, geographic and behavioural questions and location of home and workplace at resolution level of postal codes. Users are reminded weekly, via an email newsletter, to report their health status through a brief symptoms questionnaire. In our analysis, data from 2013/2014 for Italy are used. The platform in Italy is called Influweb ( www.influweb.it ). Starting from week 46 of 2013, we have collected data about ILI daily incidence in each postal code in which there are active participants. In order to provide weekly predictions, we combined web surveillance data, historical sentinel surveillance data and an epidemic stochastic generative model (GLEAM, http://www.gleamviz.org ) to predict the timing, peak and intensity of the current influenza season. Results The results of the predictions obtained with this methodology provide the following indicators of the current influenza season: -Start of the influenza season at the national level -Peak week at the national level -The highest numeric value reached at the national level -Number of weeks the epidemic remains above baseline at the national level. According to our predictions, the flu epidemic in Italy was supposed to reach the peak in the week between 3rd and 10th of February 2014. This prediction has been confirmed by the data collected by the national sentinel surveillance. The geographical trend has not been uniform, as a higher flu activity in the northern part of the country was predicted and observed. Conclusions The integration of digital surveillance data from participatory systems with the modeling approach is a powerful combination which provides a tool to make predictions, several weeks in advance, about the unfolding of the seasonal influenza in a specific country. Future work will foresee an integration with the national syndromic surveillance system responsible for preparedness and response, also in order to enhance user participation, retention and engagement. []