How to run the Shiny app

After installing the package with install.packages(“DiagnostySize”) one can run the app with

DiagnostySize::run_app()

Documentation for the DiagnostySize app

The DiagnostySize app is an application for calculating the sample size or the power in diagnostic accuracy studies. The optimal calculation approach was developed by Stark et al.

First of all the user can select an endpoint. If the user wants to calculate the sample size for a diagnostic accuracy trial, select “sample size”. Alternatively, the user can calculate the power of a diagnostic accuracy study. To do this, click on “power”. Next, select a study design. In the single-arm trial, each participant receives the experimental test and the reference standard test.

Fig.1: Screenshot shows the endpoint and design choices

The comparative design compares two tests. One test is a previously evaluated comparator test and the second test is the experimental test. When you select the comparative design, you can choose between two types of comparison design. The unpaired approach and the paired approach. In the unpaired approach, the participants are divided into two groups, one receiving the experimental test and the other receiving the comparator test. Both groups also receive the reference standard test.
In the paired design, each participant receives all three tests, the reference standard, the comparator and the experimental test. In the comparative design, you can test whether the experimental test is superior to the comparative test, or you can set a non-inferiority boundary for both endpoints (sensitivity and specificity).

Fig.2: Screenshot shows the non-inferiority limits and a tooltip with additional informations

If the user chooses a non-inferiority boundary, the alpha value entered is divided by 2 because alpha is one-sided in a non-inferiority test.

Once these choices have been made, the user must complete the form displayed. There are default values that can be easily replaced or adjusted. The arrows on the right of the input box allow the user to move up and down by 0.01 steps.

Fig.3: Screenshot shows the numeric input field with arrow to change the value.

Note: The values are processed using a decimal point as the decimal separator. If the browser language is set to German, the separator may be displayed as a comma instead.

Valid entries are between 0 and 1 in each field, except for the sample size, which is valid between 0 and 999999. Leaving an input field empty causes errors. Pleas fill out all of the input fields. If a negative value is entered, an error message is displayed in the main part of the application.
For further information, the user can hover over the input field and an additional explanation of the value will pop up. An example of this additional information is shown in screenshot 2 & 3. In screenshot 1 you can see question mark button which, when clicked, displays further explanations about the study designs.

The right hand side of the application displays the results of the calculation. These change as the design or input values are changed. Below the calculation results is a result text listing the input and output values. Also there is information about the paper published by Stark et al. These papers serve as the basis for the sample size or power calculation .