Extended Data Fig. 7: Extended patient classifier workflow schema.
From: Interpretable inflammation landscape of circulating immune cells
(a) Definition of reference and query datasets, for Scenarios 1, 2, 3, and centralized dataset (from left to right). (b) Integration of the reference dataset and mapping of the query dataset to define the patient-wise embeddings, stratified by cell type. (c) Patient classifier pipeline composed by the hyperparameter tuning of each classifier family, the selection of the best classifier family and the final evaluation of the left-out query dataset. (d) Schema of the three experiments performed. Icons created with Inkscape.
