Skip to main content
. 2022 Jan 26;18(1):e1009826. doi: 10.1371/journal.pcbi.1009826

Fig 1. The setup of the model.

Fig 1

XN×p is the input data. There are p variables and N samples. YN×q is the output data. There are q variables and N samples. EN×r is the low-dimensional nonlinear data embedding, where r is small. Clinical confounders such as age, gender, ethnicity, batch etc, form the matrix CN×s.