Skip to main content
. 2023 May 10;15(10):2696. doi: 10.3390/cancers15102696

Figure 1.

Figure 1

Workflow for development and testing the 3D genomic classifier model for prediction of ICI treatment response outcome. From EpiSwitch array screening profiles accounting for over 30 million data points, 72 top array markers were selected based on Glmnet logistic regression. Twenty-four markers qualified for translation into EpiSwitch qPCR format, of which 20 markers have passed quality control and feature reduction control on 32 patient samples. A training cohort of 77 patients was used to build a predictive classifier model based on eight qPCR markers. It was then validated on independent cohorts of 24, 128, and 51 patients.