The incidence of influenza-like symptoms has been on the rise, but computer scientists and statisticians at Carnegie Mellon University say their models suggest flu activity may be reaching its peak for the 2016-17 season.
One of the flu forecasting systems developed by Carnegie Mellon's Delphi research group shows flu activity peaking nationally this week and next, while a second system predicts the peak occurring the week of Feb. 5. The flu season continues through May.
The Stat system, which uses machine learning to make predictions based on past patterns and on current input from the U.S. Centers for Disease Control and Prevention's domestic influenza surveillance system, currently predicts activity peaking nationwide this week and next. The same system predicts flu activity for the region that includes Pennsylvania, West Virginia, Delaware, Maryland and Virginia will peak the week of Feb. 12.
A second Delphi system, called Epicast, is based on the best guesses made weekly by a group of people, who likewise receive input from the CDC's surveillance network. That system is predicting that the national peak in flu-like symptoms will occur the week of Feb. 5, with the peak for the region including Pennsylvania occurring the week of Jan. 29.
CMU's systems are part of a CDC research initiative to develop methods of accurately forecasting flu activity. For the 2015-16 flu season, the CMU systems proved to be the most accurate in their forecasts, besting 11 competing systems fielded by 10 other groups.
Roni Rosenfeld, a professor in the School of Computer Science's Machine Learning Department and Language Technologies Institute, cautioned that the forecasting models are in the research stage and are not ready to be used to make decisions about vaccination campaigns or about staffing or scheduling in health care facilities.
He and other members of the Delphi group, which includes faculty and students from CMU's machine learning, statistics, computer science and computational biology departments, anticipate they will eventually forecast flu activity in much the same way that meteorologists make weather forecasts. Tools developed for flu forecasts might also be applied to such diseases and conditions as HIV, drug resistance, Ebola, Zika and Chikungunya.
People can help the Delphi group's efforts by joining its "wisdom of crowds" forecasting system. Anyone can participate by registering on the Epicast website.
CMU's Delphi group belongs to a University of Pittsburgh-based MIDAS National Center of Excellence, a National Institutes of Health-funded network of researchers developing computational models to guide responses to disease outbreaks.