Predicting Advanced Prostate Cancer from Modeling Early Indications in Biopsy and Prostatectomy Samples via Transductive Semi-Supervised Survival Analysis

Table 7

Weights of features in models with multimodal characteristics from biopsy data.

	Cox Model		Cox Model with Transduction		SVRc		SVRc with Transduction
Feature	Original Weight	Normalized Weight	Original Weight	Normalized Weight	Original Weight	Normalized Weight	Original Weight	Normalized Weight
PSA	1.715	1.000	1.720	1.000	-60.799	-5.430	-62.229	-6.859
Dominant Biopsy Gleason Grade	0.565	0.329	0.467	0.272	-21.025	-1.878	-23.007	-2.536
Biopsy Gleason Sum	0.476	0.278	0.586	0.341	-22.943	-2.049	-22.912	-2.525
AR Expression	0.251	0.146	0.185	0.108	-18.932	-1.691	-19.903	-2.194
Nuclear Morphology	-0.228	-0.133	-0.263	-0.153	-0.866	-0.077	-0.001	0.0001
Ki67+ Area	0.885	0.516	0.793	0.461	-28.498	-2.545	-28.366	-3.127
Luminal Area	-0.171	-0.100	-0.159	-0.093	11.197	1.000	9.073	1.000
Epithelial Cells Infiltration	-0.127	-0.074	-0.161	-0.094	-2.226	-0.199	-3.040	-0.335
Glandular Size	0.088	0.052	0.019	0.011	-9.765	-0.872	-8.777	-0.967