Fig. 1: Basic concepts and terms of framing. | npj Digital Medicine

Fig. 1: Basic concepts and terms of framing.

From: The Framing of machine learning risk prediction models illustrated by evaluation of sepsis in general wards

Fig. 1: Basic concepts and terms of framing.

a Illustrates prediction time, observation window, and prediction window. Window shift (b) is the distance between each prediction. c Shows a variation of (a) where the prediction window does not begin immediately after the prediction time but instead shifts to the lead window. d Shows an example of left-aligned samples, where the model is used for predicting fixed time points, such as admission or preoperatively. e Shows an example of right-aligned samples where all data are considered until some time point before the event of interest (for the positive cases) or the end of admission (for the negative cases).

Back to article page