README.md

In muxuanliang/concordLearn: This package implments the joint-model-assisted high dimensional linear decision rule

Can a joint model assist target label prediction? Conditions and approaches

Description: The package implements the proposed method used in the simulation and application to heart failure readmission prediction in this paper.

License: MIT

Requirements

This is an R package and requires R (>= 4.0.1)

Imports: glmnet

Encoding: UTF-8

Install R package

To install the R package, run this command:

R CMD INSTALL 

Synthetic Dataset

To evaluate the proposed method by synethetic dataset, run the following commend in R:

source("test_count.R")

Example

The following example shows hwo to use the package and the main functrion cInfer. Details can be found in the help document in R.

Synethetic data generation

``` Synethetic data generation nobs <- 500 p <- 500 rate <- 0 alpha <- 0 V <- function(p, rate = 0.5){ V.matrix <- array(0, c(p,p)) for (i in 1:p){ for (j in 1:p){ V.matrix[i,j] <- rate ^ (abs(i-j)) } } V.matrix } x <- mgcv::rmvn(nobs, rep(0, times=p), V(p, rate)) beta_true <- c(1,-1,1,-1, rep(0, times=p-4)) beta_modify <- c(1,1,1,1, rep(0, times=p-4)) mix <- rbinom(nobs, 1, 1-alpha) U <- sapply(pnorm(x%*%beta_true), function(t){ v <- rbinom(1, 4, t) if (v==3){ v.add <- rbinom(1, 1, alpha) v <- v+v.add } else if (v==4){ v.add <- rbinom(1, 1, alpha) v <- v-v.add } v }) y.cutoff <- list(U>0, U>3)

### Fit using the proposed method
``` fit using proposed method
fit <- cInfer(x, y=y.cutoff, y_refit = list(y.cutoff[[1]]), weight = c(1, 1, 1, 1, rep(1, times= p-4)), lossType = 'logistic', tol = 1e-3, parallel = FALSE))

Results

Our model achieves the following performance in the application to heart failure readmission prediction:

| Model name | Accuracy | Se | | ------------------ |---------------- | -------------- | | Proposed | 76.1% | 7.18*1e-4 |

muxuanliang/concordLearn documentation built on Dec. 21, 2021, 11:04 p.m.