Differential Misclassification considering Specificity and Sensitivity (DMSS) model uses binary regression models to fit the covariate-related sensitivity and specificity simultaneously for the observed response.
You can install the development version of DMSS from GitHub with:
# install.packages("devtools")
devtools::install_github("AnqiWang2021/DMSS")This is a basic example which shows you how to solve a common problem:
library(DMSS)
library(rje)
# Generate the predictor variable and covariates
X = rnorm(10000)
L = rnorm(10000)
V = rnorm(10000)
# Generate the covariate-related sensitivity and specificity
theta = c(-2,3)
p_Dtrue = expit(cbind(1,X)%*% theta)
beta = c(0.45,0.5)
sensitivity = expit(cbind(1,V)%*% beta)
gamma = c(4.5,1)
specificity = expit(cbind(1,L)%*% gamma)
# Generate the observed outcome variable
p_Dstar = (1-specificity)+(specificity+sensitivity -1)*p_Dtrue
Dstar = rbinom(10000,1,p_Dstar)
# initial value
start = c(-1,1,0,0,3,0)
result = loglik_EM (Dstar,X,V,L,start,tol = 1e-8, maxit = 1000)