multisite.accuracy GUI

Please ensure that your CSV file includes the following columns:

• [for accuracy estimates other than hazard ratio] "y": actual value of the variable that you want to predict. For AUC (area under the ROC curve) accuracy estimates, it must be binary. For BAC (balanced accuracy) accuracy estimates, it must be either binary or a factor. For correlation and MSE (mean squared error) accuracy estimates, it must be a numeric.
• [for HR (hazard ratio) accuracy estimate] "time" and "status": actual time and status (1 for occurrence, 0 for censoring) of the event you want to predict.
• "y.pred": predicted value of the variable. For AUC, correlation, HR, and MSE accuracy estimates, it must be numeric. For BAC accuracy estimates, it must be either binary or a factor with the same levels as "y".
• "site": Name of the site for each observation.

Choose a CSV file:

Please choose the accuracy estimate, which may be:

• AUC (area under the ROC curve), when "y" is a binary variable and "y.pred" is a continous variable.
• BAC (balanced accuracy), along with sensitivity and specificity when both "y" and "y.pred" are binary variables, or along with the sensitivity to detect each group when both "y" and "y.pred" are factors.
• Correlation or MSE (mean squared error), when both "y" and "y.pred" are numeric variables.
• HR (hazard ratio), when instead of "y" there is "time" and "status", and "y.pred" is a numeric variable.

Estimate: AUC BAC Correlation Mean Squared Error Hazard Ratio

We recommend not changing them unless there are reasons to.

Site Method: Covariate (default) Meta-analysis None

Use mixed-effects models (if available):

Minimum site sample size:

Please click below to estimate the multisite accuracy:

Estimate the accuracy!

Estimated Accuracy:

[The results will be here]