add pkgdown build to github actions

Author: egillax

.Rbuildignore

@@ -1,2 +1,3 @@
 ^.*\.Rproj$
 ^\.Rproj\.user$
+^\.github$

.github/.gitignore

@@ -0,0 +1 @@
+*.html

.github/workflows/pkgdown.yaml

@@ -0,0 +1,35 @@
+# Workflow derived from https://github.com/r-lib/actions/tree/master/examples
+# Need help debugging build failures? Start at https://github.com/r-lib/actions#where-to-find-help
+on:
+  push:
+    branches: [main, develop]
+  release:
+    types: [published]
+  workflow_dispatch:
+
+name: pkgdown
+
+jobs:
+  pkgdown:
+    runs-on: ubuntu-latest
+    env:
+      GITHUB_PAT: ${{ secrets.GITHUB_TOKEN }}
+    steps:
+      - uses: actions/checkout@v2
+
+      - uses: r-lib/actions/setup-pandoc@v1
+
+      - uses: r-lib/actions/setup-r@v1
+        with:
+          use-public-rspm: true
+
+      - uses: r-lib/actions/setup-r-dependencies@v1
+        with:
+          extra-packages: pkgdown
+          needs: website
+
+      - name: Deploy package
+        run: |
+          git config --local user.name "$GITHUB_ACTOR"
+          git config --local user.email "[email protected]"
+          Rscript -e 'pkgdown::deploy_to_branch(new_process = FALSE)'

DeepPatientLevelPrediction.Rproj

@@ -14,6 +14,7 @@ LaTeX: pdfLaTeX
 
 BuildType: Package
 PackageUseDevtools: Yes
+PackageCleanBeforeInstall: Yes
 PackageInstallArgs: --no-multiarch --with-keep.source
 PackageBuildArgs: --compact-vignettes=both
 PackageCheckArgs: --as-cran

vignettes/BuildingDeepModels.Rmd

@@ -23,11 +23,13 @@ output:
     number_sections: yes
     toc: yes
 ---
+
+```{=html}
 <!--
 %\VignetteEngine{knitr}
 %\VignetteIndexEntry{Building Deep Learning Models}
 -->
-  
+```
 ```{r setup, include=FALSE}
 knitr::opts_chunk$set(echo = TRUE)
 ```
@@ -36,30 +38,30 @@ knitr::opts_chunk$set(echo = TRUE)
 
 Patient level prediction aims to use historic data to learn a function between an input (a patient's features such as age/gender/comorbidities at index) and an output (whether the patient experienced an outcome during some time-at-risk). Deep learning is example of the the current state-of-the-art classifiers that can be implemented to learn the function between inputs and outputs.
 
-Deep Learning models are widely used to automatically learn high-level feature representations from the data, and have achieved remarkable results in image processing,  speech recognition and computational biology. Recently, interesting results have been shown using large observational healthcare data (e.g., electronic healthcare data or claims data), but more extensive research is needed to assess the power of Deep Learning in this domain.
+Deep Learning models are widely used to automatically learn high-level feature representations from the data, and have achieved remarkable results in image processing, speech recognition and computational biology. Recently, interesting results have been shown using large observational healthcare data (e.g., electronic healthcare data or claims data), but more extensive research is needed to assess the power of Deep Learning in this domain.
 
-This vignette describes how you can use the Observational Health Data Sciences and Informatics (OHDSI) [`PatientLevelPrediction`](http://github.com/OHDSI/PatientLevelPrediction) package and [`DeepPatientLevelPrediction`](http://github.com/OHDSI/DeepPatientLevelPrediction) package to build Deep Learning models. This vignette assumes you have read and are comfortable with building patient level prediction models as described in the  [`BuildingPredictiveModels` vignette](https://github.com/OHDSI/PatientLevelPrediction/blob/main/inst/doc/BuildingPredictiveModels.pdf). Furthermore, this vignette assumes you are familiar with Deep Learning methods.
+This vignette describes how you can use the Observational Health Data Sciences and Informatics (OHDSI) [`PatientLevelPrediction`](http://github.com/OHDSI/PatientLevelPrediction) package and [`DeepPatientLevelPrediction`](http://github.com/OHDSI/DeepPatientLevelPrediction) package to build Deep Learning models. This vignette assumes you have read and are comfortable with building patient level prediction models as described in the [`BuildingPredictiveModels` vignette](https://github.com/OHDSI/PatientLevelPrediction/blob/main/inst/doc/BuildingPredictiveModels.pdf). Furthermore, this vignette assumes you are familiar with Deep Learning methods.
 
 # Background
 
-Deep Learning models are build by stacking an often large number of neural network layers that perform feature engineering steps, e.g embedding, and are collapsed in a final softmax layer (basically a logistic regression layer). These algorithms need a lot of data to converge to a good representation, but currently the sizes of the large observational healthcare databases are growing fast which would make Deep Learning an interesting approach to test within OHDSI's [Patient-Level Prediction Framework](https://academic.oup.com/jamia/article/25/8/969/4989437). The current implementation allows us to perform research at scale on the value and limitations of Deep Learning using observational healthcare data. 
+Deep Learning models are build by stacking an often large number of neural network layers that perform feature engineering steps, e.g embedding, and are collapsed in a final softmax layer (basically a logistic regression layer). These algorithms need a lot of data to converge to a good representation, but currently the sizes of the large observational healthcare databases are growing fast which would make Deep Learning an interesting approach to test within OHDSI's [Patient-Level Prediction Framework](https://academic.oup.com/jamia/article/25/8/969/4989437). The current implementation allows us to perform research at scale on the value and limitations of Deep Learning using observational healthcare data.
 
-In the package we have used [torch](https://cran.r-project.org/web/packages/torch/index.html) and [tabnet](https://cran.r-project.org/web/packages/tabnet/index.html)  but we invite the community to add other backends.  
+In the package we have used [torch](https://cran.r-project.org/web/packages/torch/index.html) and [tabnet](https://cran.r-project.org/web/packages/tabnet/index.html) but we invite the community to add other backends.
 
-Many network architectures have recently been proposed and we have implemented a number of them, however, this list will grow in the near future. It is important to understand that some of these architectures require a 2D data matrix, i.e. |patient|x|feature|, and others use a 3D data matrix |patient|x|feature|x|time|. The [FeatureExtraction Package](www.github.com\ohdsi\FeatureExtraction) has been extended to enable the extraction of both data formats as will be described with examples below.
+Many network architectures have recently been proposed and we have implemented a number of them, however, this list will grow in the near future. It is important to understand that some of these architectures require a 2D data matrix, i.e. \|patient\|x\|feature\|, and others use a 3D data matrix \|patient\|x\|feature\|x\|time\|. The [FeatureExtraction Package](www.github.com\ohdsi\FeatureExtraction) has been extended to enable the extraction of both data formats as will be described with examples below.
 
 Note that training Deep Learning models is computationally intensive, our implementation therefore supports both GPU and CPU. It will automatically check whether there is GPU or not in your computer. A GPU is highly recommended for Deep Learning!
 
 # Non-Temporal Architectures
+
 We implemented the following non-temporal (2D data matrix) architectures:
 
-	1) ...
+    1) ...
 
 For the above two methods, we implemented support for a stacked autoencoder and a variational autoencoder to reduce the feature dimension as a first step. These autoencoders learn efficient data encodings in an unsupervised manner by stacking multiple layers in a neural network. Compared to the standard implementations of LR and MLP these implementations can use the GPU power to speed up the gradient descent approach in the back propagation to optimize the weights of the classifier.
 
 ##Example
 
-
 # Acknowledgments
 
 Considerable work has been dedicated to provide the `DeepPatientLevelPrediction` package.
@@ -69,5 +71,5 @@ citation("PatientLevelPrediction")
 ```
 
 **Please reference this paper if you use the PLP Package in your work:**
-  
-[Reps JM, Schuemie MJ, Suchard MA, Ryan PB, Rijnbeek PR. Design and implementation of a standardized framework to generate and evaluate patient-level prediction models using observational healthcare data. J Am Med Inform Assoc. 2018;25(8):969-975.](http://dx.doi.org/10.1093/jamia/ocy032)
+
+[Reps JM, Schuemie MJ, Suchard MA, Ryan PB, Rijnbeek PR. Design and implementation of a standardized framework to generate and evaluate patient-level prediction models using observational healthcare data. J Am Med Inform Assoc. 2018;25(8):969-975.](http://dx.doi.org/10.1093/jamia/ocy032)