Returning users, click on tabs above to begin making selections OR download files below and return via email
Word document of this week's Worksheet with epidemic curves
INFORMED CONSENT INFORMATION (DRAFT)
1. The purpose of the study is to compare the effectiveness of a Wisdom of Crowds approach to creating ensemble models through crowdsourced voting compared to the effectiveness of taking a simple average.
2. I do not foresee any risks or discomforts to the crowd members other than the time required. Making selections for is expected to take 5 to 10 minutes.
3. Hopefully, crowd members find selecting COVID-19 models somewhat enjoyable and learn about COVID-19 and COVID-19 models during the process.
4. Through the Epidemic Prediction Initiative, CDC leads several forecasting challenges. More information is available at https://predict.cdc.gov
5. Crowd members may contact Jeff Morgan at 19morgan@cua.edu if they have any concerns. The website is hosted by HostGator and some additional levels of protection were purchased. Details available upon request and will be added in future iterations.
6. Crowd members participation is strictly voluntary. Crowd members may submit picks as often as they wish
7. There is no cost to participate.
8. A prize structure has not yet been determined, but random drawings for cash is expected
Introduction
Thank you for your interest in this Wisdom of Crowds Approach to forming an Ensemble Model for COVID-19 based on compiling votes for individual models.
Your votes will be compiled with other voters to form a wisdom of the crowd.
It is currently planned that Votes will be collected until the end of October.
Please make selections at the National Level.
Because symptoms for COVID-19 are also symptoms for Influenza Like Illness, one of CDC's metrics for measuring the intensity of the COVID-19 pandemic is based on measuring Inluenza Like Illness
The Epidemic Curve for Influenza Like Illness Percentage shows the trajectory of the influenza seasons for the the 2009 H1N1 Pandemic and most recent seasons. Each model provides a probability distribution for each "target". The "targets" are the wILI% for one week ahead, two weeks ahead, three weeks ahead, four weeks ahead; season onset; season peak week; and season peak wILI%. More information is available in the links below.
Links to Background Information
Wisdom of Crowds Background
CDC Epidemic Prediction Initiative
Explanation of Probability Distributions
CDC paper in the American Journal of Epidemiology on the classification of the severity of influenza seasons and pandemics by Biggerstaff et.al."
CDC website describing how CDC classifies the severity of influenza seasons and pandemics
CDC website describing the U.S. Influenza Surveillance System Presentation to CUA BE 499
Poster for CSTE/CDC FluSight Seasonal Influenza Forecasting Workshop
Instructions for Voting via Online Selections
(1) Navigate to each page at the top of this page or use the links below:
(2) Look at the epidemiological curves and table to view that area's wILI% for the current season and previous seasons. Click on the image to open the curve(s)/table in another tab(s)/window(s) if desired.
(3) For each area and target (e.g., season onset, one week ahead), look at the probability distribution of each model and select the best model.
(4) Click the submit button.
(5) Navigate to other pages as desired.
Complete U.S. National vote online.Instructions for voting via emailing an MS Word file
(1) Download the Worksheet(s) and Ballot(s).
(2) Use the Worksheet to view the wILI% of the current season and previous seasons.
(3) Select the best model for each target and area
(4) Save file and email to 19morgan@cua.edu.
Word version of this week's Ballot due Mar 26