Epidemiological forecasting is critically needed for decision making by public health officials, commercial and non-commercial institutions, and the general public. The DELPHI group focuses on developing the technological capability of epi-forecasting, and its role in decision making, both public and private. Our long term vision is to make epidemiological forecasting as universally accepted and useful as weather forecasting is today.
Geographically detailed real-time estimates (nowcasts) of Influenza-Like-Illness in the U.S.: ILI-Nearby
Weekly forecasts of Influenza-Like-Illness nationally and in 10 U.S. regions, by different forecasting methods: Delphi Flu Forecasting
‘Wisdom-of-Crowds’ system for crowdsourced forecasting of the flu: Epicast
Publicly Available Tools
Epidemiological time series visualizer: EpiVis
Visual comparative score analysis of submissions to CDC's Predict the Flu Challenge (provide your own score files): FluScores
Source codes are freely available on GitHub.
We have participated, and have done very well, in all epidemiological forecasting challenges organized by the U.S. government to date:
Ongoing Research Projects
In-season epidemiological forecasting via statistical machine learning
Our 2015 PLoS Computational Biology paper, Flexible Modeling of Epidemics with an Empirical Bayes Framework, describes the Empirical Bayes methodology we developed and its application to in-season forecasting of flu in the U.S. See also our 2014 PLOS Neglected Tropical Diseases paper on applying this methodology to in-season forecasting of dengue in Brazil. A manunscript describing our most ecent ensemble system (#1 performer in the 2015—2016 Predict the Flu challenge) is under preparation.
Another one of our goals is to establish a strong baseline forecast to which data-driven forecasts can be compared. For chikungunya (in 2015) and influenza (ongoing since 2014), we achieve this baseline through periodic collection of simple, subjective forecasts from the general public, using a graphical interface we created specifically for this purpose. With these user-submitted assessments we create aggregate, probabilistic weekly forecasts, both as a performance baseline for comparison with data-driven methods and also as a serious attempt to forecast the spread of these diseases. A manuscript describing the Epicast system has recently been accepted for publication. See also this section of David Farrow's thesis.
Join in! We are always looking for more participants, so please check out Epicast (http://epicast.org) - and feel free to invite others to do so as well. No special skills or background are required, and the top epicasters are featured on the leaderboards!
Nowcasting by sensor fusion
Description coming soon...
Sangwon (Justin) Hyun
Pu "Paul" Liang
Zirui "Edward" Wang
We are part of a large University of Pittsburgh-based MIDAS National Center of Excellence, which includes epidemiologists, virologists, public health experts, medical doctors specialized in infectious disease, legal and economic experts, and computationalists.
Roni's Proposal for Standardized Evaluation of Epidemiological Models whitepaper
Roni's Predicting The Predictable presentation
We maintain a list of publications relating to epi-forecasting here.––>