The R package QDRS provides functions for calculating quantitative disease risk scores, with application to Eletronic Health Records.
You can install the development version of QDRS from Github:
# install.packages("devtools")
devtools::install_github("danqingxu/QDRS")
library(QDRS)
# EHR demo data
# sample.set: matrix of binary features.
# sample.group: a grouping factor with two levels "Case" and "Control".
# training: an index vector that indicates training rows.
sample.set = EHR$sample.set
sample.group = EHR$sample.group
training = EHR$training
# Compute QDRSs
PheRS.res = PheRS(X = sample.set, group = sample.group)
Eigen.res = eigen.score(X = sample.set, training = training, scale = TRUE)
PC.res = PC(X = sample.set, group = sample.group, training = training, scale = TRUE, pc.num = 1:2)
LPC.res = LPC(X = sample.set, group = sample.group, training = training, scale = TRUE)
LVS.res = LVS.score(X = sample.set, Y = NULL, family = "binomial", starting.choice = "random", p.seed = 124)
NMF1.res = NMF1(X = sample.set)
# Assess the performance
pairwise.wilcox(x = LPC.res$scores, g = sample.group)
pairwise.auc(x = LPC.res$scores, g = sample.group)
pairwise.upr(x = LPC.res$scores, g = sample.group)
https://github.com/danqingxu/QDRS_Documentation
This package is free and open source software, licensed under GPL (>=2).