select_top_features: Select top radiomic features according to a given statistics

In cgplab/RadAR: Radiomic Analysis with R

Description

This function create a summary table of most significant features based on previously computed statistics (wilcox, AUC, concordance index or cox regression)

Usage

select_top_features(
  rdr = NULL,
  which_statistics = "wilcox",
  thr_pvalue = 0.05,
  thr_auc = 0.8,
  thr_concordance = 0.1,
  thr_cox_zvalue = 3,
  write_to = NULL
)

Arguments

rdr	A RadAR object (class SummarizedExperiment).
which_statistics	(character) Select top features based on one of the following pre-computed statistics: "wilcox", "AUC", "concordance" or "cox".
thr_pvalue	(numeric) P-value threshold to identify statistically significant features from wilcox. It should be in the range (0, 1].
thr_auc	(numeric) AUC threshold to identify statistically significant features from AUC It should be in the range (0.5, 1].
thr_concordance	(numeric) Threshold to identify statistically significant features from concordance analysis. It should be in the range (0, .5].
thr_cox_zvalue	(numeric) Z threshold (Wald statistics) to identify statistically significant features from cox regression analysis. It should be in the range (0, inf).
write_to	(character) If specified, filename to output top statistically significant features (tab-delimited).

Value

A table of class tibble reporting top features.

Author(s)

Matteo Benelli ([e-mail])

cgplab/RadAR documentation built on Nov. 10, 2021, 1:32 a.m.