Research Article
Early Prediction of Sepsis Based on Machine Learning Algorithm
Table 5
Time window structure of different variables.
| |||||||||||||||||||
Note: (1) when the length of the time window is 2 h, if the current time t < 2, the feature value is nan, and the missing position is filled with the median of the column data; (2) when the length of time windows is 12 h/24 h, if the current time t < 12/24, the time window will change into time from ICULOS = 1 to the current time ICULOS = t to construct features. |