Table 1. Overview of train-test split scenarios.

Train-test split scenario	Explanation	Visualization
Group 5-fold cross-validation	Separation of data points into five bins of approximately equal size, with the condition that each patient’s data are represented in exactly one bin, that is, either in the training set or the test set, but not both. Train on all but one bin, test on the remaining bin. Repetition of the procedure until each bin has been used once as a test bin (5-fold cross-validation).	a
Leave-one-subject-out	Train on data from all but one patient. Test on data from the one left-out patient. The procedure was repeated until each subject was used in the test arm (in our study N=22).
Chronological split	Train on the chronologically first 50% of data, test on the last 50%.
Odd-even split	Odd assessment points were assigned to the training set, even assessment points to the test set. Then the implementation of a 2-fold cross-validation.
Random split	Data points were randomly assigned to either train or test sets. This was repeated ten times with a 2-fold cross-validation calculated in each repeated run.

^aFor visualizations: squares represent data bins in the first row and individual patients in the remaining rows; circles represent individual data points. P: patient.