Physicists in Italy have begun analyzing data from a new Web-based project that seeks to model how flu spreads through a population.

The project, known as Influweb, involved some 2000 ordinary Italians replying over the last six months to a weekly e-mailed questionnaire about their state of health and current geographical location. The project will be able to pin down the spread of flu in real time and with a spatial resolution on the level of a person’s postcode.

The spread of flu is traditionally monitored using information provided by doctors after they have seen ill patients. One problem with this approach is that it can take over a week for the data to be processed. Moreover, such methods do not sample enough people.

Most don’t see a doctor

“We found that 90% of people with symptoms never see a doctor,” explains physicist Daniela Paolotti from the Institute for Scientific Interchange (ISI) in Turin, who is a joint co-ordinator of Influweb along with Vittoria Colizza.

What the ISI researchers hope to obtain is much more accurate data that will allow them to test and refine their theoretical models of the spread of flu and other epidemics. This should, in principle, allow them to make better predictions of how widespread a particular outbreak will be and when it will peak.

Traditional monitoring techniques have so few data that they only become statistically relevant if summed over a large geographical area. The Influweb team now has six months worth of data, including information on groups such as pensioners and children.

Alerting hospitals

Influweb is inspired by similar Web-based projects that have been running in Belgium and the Netherlands since 2003 and in Portugal since 2005. The former scheme, which has information on 60,000 people, has been able to predict a week earlier than other techniques when a particular epidemic will peak. The Influweb project will, however, try to exploit that breathing space by alerting hospitals of an imminent epidemic so that they can, for example, put extra ambulances on stand-by.

Although Paolotti has not obtained enough data this winter to make full predictions, she is optimistic for next year. “We are confident that next winter will yield even more participation and that we will gain predictive capacity,” she says.

The ISI researchers are also co-ordinating a €5m four-year European-wide extension of Influweb, known as Epiwork. Funded by the European Union and co-ordinated by Alessandro Vespignani from the ISI, Epiwork began in February and involves 12 organizations from seven European countries and Israel.

“I believe this project is a cutting-edge approach to data collection,” says Zoltán Toroczkai, a physicist at the University of Notre Dame in Indiana, who is also involved in modelling epidemics. “It will be extremely useful in tackling a problem that has not really been exploited in the current modelling, namely the dynamical coupling between people’s mobility in geographical space and the spread of an epidemic.”