Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/genentech/jmpost
https://github.com/genentech/jmpost
Last synced: 1 day ago
JSON representation
- Host: GitHub
- URL: https://github.com/genentech/jmpost
- Owner: Genentech
- Created: 2022-06-01T20:43:02.000Z (over 2 years ago)
- Default Branch: main
- Last Pushed: 2024-09-17T10:29:54.000Z (10 days ago)
- Last Synced: 2024-09-18T11:13:48.484Z (9 days ago)
- Language: R
- Homepage: https://genentech.github.io/jmpost/
- Size: 40.3 MB
- Stars: 17
- Watchers: 6
- Forks: 4
- Open Issues: 56
-
Metadata Files:
- Readme: README.Rmd
- Changelog: NEWS.md
Awesome Lists containing this project
- top-pharma50 - **genentech/jmpost** - 06-07 16:30:15 | (Ranked by starred repositories)
- top-pharma50 - **genentech/jmpost** - 06-07 16:30:15 | (Ranked by starred repositories)
README
---
output: github_document
editor_options:
chunk_output_type: console
---
# jmpost 
```{r, echo = FALSE}
knitr::opts_chunk$set(
collapse = TRUE,
comment = "#>",
fig.path = "README-"
)
```
[](https://www.repostatus.org/#wip)
[](https://raw.githubusercontent.com/genentech/jmpost/_xml_coverage_reports/data/main/coverage.xml)
\
The goal of the `jmpost` package is to fit joint models involving:
1. a parametric time-to-event sub-model,
1. a nonlinear (or linear) mixed-effect sub-model, describing individual time profiles (_i.e._ trajectories) for a continuous marker,
1. a link function (_a.k.a._ association term).
More specifically, the model implemented in this package utilizes a modelling framework described previously **[1-3]** to link overall survival to tumour size data in oncology clinical trials.
**[1]** [Tardivon _et al._ Association between tumour size kinetics and survival in patients with urothelial carcinoma treated with atezolizumab: Implications for patient follow-up. _Clin Pharm Ther_, 2019](https://doi.org/10.1002/cpt.1450).
**[2]** [Kerioui _et al._ Bayesian inference using Hamiltonian Monte-Carlo algorithm for nonlinear joint modelling in the context of cancer immunotherapy. _Stat in Med_, 2020](https://doi.org/10.1002/sim.8756).
**[3]** [Kerioui _et al._ Modelling the association between biomarkers and clinical outcome: An introduction to nonlinear joint models. _Br J Clin Pharm_, 2022](https://doi.org/10.1111/bcp.15200).
The models are implemented in [STAN](https://mc-stan.org/), and the package provides a flexible user interface.
Please reach out to us via issues or email (see the `DESCRIPTION` file) if you have comments or questions or would like to get involved in the ongoing development, thank you!
## Installation
**GitHub**
You can install the current development version from GitHub with:
```{r gh-installation, eval = FALSE}
if (!require("remotes")) {
install.packages("remotes")
}
remotes::install_github("genentech/jmpost")
```
Please note that this package requires [`cmdstanr`](https://mc-stan.org/cmdstanr/).
**CRAN**
This package has not been published to CRAN yet.
## Getting Started
See also the [model fitting](https://genentech.github.io/jmpost/main/articles/model_fitting.html)
vignette for more details. Here we present a very basic example here.
First we simulate a data set. In practice you want to follow a similar structure
of the input data and use `DataJoint()` to bring it into the right format.
```{r sim_data}
library(jmpost)
set.seed(321)
sim_data <- SimJointData(
design = list(
SimGroup(50, "Arm-A", "Study-X"),
SimGroup(50, "Arm-B", "Study-X")
),
longitudinal = SimLongitudinalRandomSlope(
times = c(1, 50, 100, 150, 200, 250, 300),
),
survival = SimSurvivalWeibullPH(
lambda = 1 / 300,
gamma = 0.97
)
)
joint_data <- DataJoint(
subject = DataSubject(
data = sim_data@survival,
subject = "subject",
arm = "arm",
study = "study"
),
survival = DataSurvival(
data = sim_data@survival,
formula = Surv(time, event) ~ cov_cat + cov_cont
),
longitudinal = DataLongitudinal(
data = sim_data@longitudinal,
formula = sld ~ time,
threshold = 5
)
)
```
Then we specify the joint model, here we use a Generalized Stein-Fojo model
for the longitudinal part, and a Weibull proportional hazards model for the
survival part. The longitudinal model impacts the hazard via a term for the
derivative and another term for the time-to-growth.
```{r model_spec}
joint_model <- JointModel(
longitudinal = LongitudinalGSF(),
survival = SurvivalWeibullPH(),
link = Link(
linkDSLD(),
linkTTG()
)
)
```
Finally we can sample the parameters via MCMC from the underlying Stan model.
Note that in a real application you will choose more warm up and sampling iterations.
```{r sample_model, eval = FALSE}
mcmc_results <- sampleStanModel(
joint_model,
data = joint_data,
iter_sampling = 100,
iter_warmup = 100,
chains = 1,
parallel_chains = 1
)
```
## Citing `jmpost`
To cite `jmpost` please see [here](https://genentech.github.io/jmpost/main/authors.html#citation).