Random forest results
datasetposted on 10.12.2021, 10:36 authored by Olivia RifaiOlivia Rifai
Random forest crosstable and feature results based on digital pathology features. Classifications were conducted between disease (C9-ALS) and control "conditions" overall, divided by marker (CD68/FUS/GFAP/Iba1/TDP43), brain region (BA), matter type (GM/VM), or vascular adjacency (VA). Classifications for subregions are shown as "marker-region", "marker-region-matter" and "all variables". Crosstables show how many images the model correctly classified.
"Features" files show which features, when left out, reduced or improved the ability of the model to accurately classify images.
"Impairment", "duration" and "ALSFRS" files show the inability of the model to accurately classify between these condition types (as opposed to disease versus control).