About the App

This web app enables researchers to upload GPAQ values and download predicted values which are calibrated to more accurately represent what would have been recorded via ActiGraph. Data must be uploaded as a text file; various delimiters may be used. Any rows with missing values will be removed. Visualizations are provided, and the calibrated values can be downloaded as part of the original data set.

Models Used

The calibration model was constructed using data collected by the Active Ottumwa project (https://www.public-health.uiowa.edu/prc/active-ottumwa/), a community-based research project by the University of Iowa Prevention Research Center, funded by the Centers for Disease Control and Prevention. A number of linear and non-linear relationships between GPAQ scores and corresponding ActiGraph data were investigated. AIC was used to select the best model from a training set, and the resulting calibration model was validated on a testing set from the same population as well as a testing set on a different population. Please see the following paper for more details

Metcalf, K., Baquero, B., Laroche, H., Coronado Garcia, M., Francis, S., Janz, K., and Sewell, D.K. (2017+),"Calibration of the Global Physical Activity Questionnaire to Accelerometry Measured Physical Activity and Sedentary Behavior in a Random Sample of Rural Iowans," Working paper.

If this web app is used, we request that the above work be cited.

How the App Works

1. Use the "Upload File" tab to upload your data in text format. Comma, tab, or semicolon delimited files accepted. Your uploaded data will be shown for you to confirm that it has been read in correctly.

2. The "Predictions" tab allows users to download the calibrated data. First, use the drop down menus to match the required variables. If you have ActiGraph data with which you would like to compare the predictions, select the appropriate box. To download the predictions, select the appropriate button at the bottom of the page.

3. Visualizations are also provided on the "Predictions" tab. The first plot is a Bland-Altman plot, giving the GPAQ values minus the calibrated values vs. the average of the GPAQ and calibrated values. The second plot (only when ActiGraph data is present) shows the ActiGraph values vs. both the GPAQ and calibrated values. Each observation thus corresponds to a pair of points which are connected by a gray segment. The diagonal black line is given by y=x, and is where we desire the calibrated values to fall.

4. If ActiGraph data is present, ratios of the mean squared predictive errors (MSPE) are given, where values less than one imply the calibrated values contain less error than the GPAQ values. For example, if you get a MSPE ratio of 0.12 for the sedentary physical activity your can say that calibrated sedentary data reduced the error of your sedentary GPAQ data by 88%.

Funding

This research was supported by Cooperative Agreement Number 1-U48DP001902-01 from the Centers for Disease Control and Prevention. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Questions

Any questions or problems should be sent to Daniel Sewell at daniel-sewell AT uiowa DOT edu