ppcm: Predictive Partly Conditional Model

The $\mathtt{ppcm}$ package aims to provide user-friendly functions to implement a predictive partly conditional model to characterize disease progression at time $t$ with longitudinal ordinal outcomes in the presence of time-dependent covariates at time $s$ ($s < t$) and time-dependent effects.

Installation

You can install the $\mathtt{ppcm}$ package from GitHub with:

# install.packages("devtools")
devtools::install_github("liud4/ppcm")

Import Dataset

example.df <- readr::read_csv(system.file("extdata", "example_data.csv", package = "ppcm"))
head(example.df)

## # A tibble: 6 × 6
##      id  time outcome    X1     X2   Z_ij
##   <dbl> <dbl>   <dbl> <dbl>  <dbl>  <dbl>
## 1     1     0       0     1 -0.561 -2.36 
## 2     1     1       0     1 -0.561 -2.26 
## 3     1     2       0     1 -0.561 -0.245
## 4     1     3       0     1 -0.561 -0.779
## 5     1     4       1     1 -0.561 -2.85 
## 6     2     0       1     0  0.120 18.6

Implementation

ppcm.fit <- ppcm::ppcm(
  data = example.df,
  id.var = "id",
  outcome.var = "outcome",
  time.var = "time",
  cov.ti = c("X1", "X2"),
  cov.td = "Z_ij",
  method = "ppcm",
  baseline.only = FALSE,
  fixed.window.length = 3,
  int.width = NULL,
  bandwidth = NULL,
  width.subset = FALSE
)

ppcm.res <- as.data.frame(rbind(round(ppcm.fit$est, digits = 3),
                  round(ppcm.fit$se, digits = 3)))
rownames(ppcm.res) <- c("Est.", "S.E.")

ppcm.res %>% knitr::kable(., format = "pipe", padding = 2)

	X1	X2	Z_ij
Est.	0.468	-0.039	0.510
S.E.	0.230	0.232	0.035