Are we there yet? A machine learning architecture to predict organotropic metastases

BMC Medical Genomics

Table 1 Average model metrics by cancer

TCGA-Project	Avg. Precision	Avg. Recall	Avg. F-Measure	Avg. Model accuracy
BLCA	0.93	0.87	0.89	0.90
BRCA	0.82	0.80	0.81	0.81
COADREAD	0.76	0.76	0.76	0.75
ESCA	0.77	0.81	0.79	0.81
HNSC	0.86	0.85	0.85	0.86
KIRC	0.93	0.95	0.94	0.95
KIRP	0.87	0.89	0.88	0.89
LIHC	0.95	0.91	0.93	0.93
LUAC	0.76	0.75	0.75	0.75
LUSC	0.65	0.67	0.66	0.67
PAAD	0.75	0.77	0.76	0.77
PRAD	0.88	0.87	0.86	0.87
SARC	0.70	0.75	0.72	0.75
SKCM	0.73	0.79	0.76	0.79
STAD	0.73	0.74	0.74	0.74
THCA	0.61	0.61	0.61	0.61

Displayed are the cumulative model performance metrics aggregating all locations for each cancer type. The cancers are labeled with their four letter TCGA code. Model metrics reported right to left were classification precision, classification recall, classification F-Measure and classification accuracy. Model performance variance and standard deviation are reported in the Additional file 1. Positive and Negative class specific performance reported in Additional file 1: data tables

ISSN: 1755-8794