Figure 6
From: Bayesian probit regression model for the diagnosis of pulmonary fibrosis: proof-of-principle
Validation tests. Each sample of the GSE10667 cohort is assigned a probability of IPF. Cutoffs were determined by calculating the Youden index. The true phenotype of each sample is indicated in color (15 normals [blue] and 31 IPF [red]). (A) The All IPF signature is used to assign IPF probability. (B) The IPF Biopsy signature is used to assign IPF probability. (C) The IPF Explant signature is used to assign IPF probability.