What is it about?<\/h3>\n\n\n\n
The story behind this app comes from a real application inside departments evaluating clinical studies in diagnostics.<\/p>\n\n\n\n
Due to fast recruiting for clinical studies the patient cohorts seemed to be inhomogenic. Therefore a researcher and a clinical study statistician wanted to find out, by which parameter they can find patients, that do not seem to fit their desired class. Maybe there was a mistake in the labeling of a patient\u2019s disease status? Maybe one measurement or two measurements can be used to easily find such patients?<\/p>\n\n\n\n
The example data used here is real data from a 1990\u2019s study, known as the biopsy data set, also hosted on UCI ML data repository<\/a>. The app that should be build inside this tutorial is preliminary and was especially build for the tutorial. Pieces of it were applied in real world biostatistical applications.<\/p>\n\n\n\n Please start forking the tutorial at: https:\/\/github.com\/zappingseb\/biopharma-app<\/a><\/p>\n\n\n\n and afterwards run the installation of dependencies in your R session:<\/p>\n\n\n if you have all the packages installed you will have one file to work on. This file is This file we will use to insert the right visualizations and the right table for the researcher to fullfil the task named above. To check what the app will finally look like you can already perform The app already contains:<\/p>\n\n\n\n A A It already contains a function to provide you with the input data sets In this tutorial we will go through steps 1\u20134 to enable building the app<\/p>\n\n\n\n The patients inside the data set were obtained from the University of Wisconsin Hospitals, Madison from Dr. William H. Wolberg. He assessed biopsies of breast tumours for 699 patients<\/a> up to 15 July 1992; each of nine attributes has been scored on a scale of 1 to 10, and the outcome is also known. There are 699 rows and 11 columns.<\/p>\n\n\n\n The data set can be called by the variable The and see<\/p>\n\n\n\n We now need to build the And you\u2019re done.<\/p>\n\n\n\nStarting point<\/h3>\n\n\n\n
\ninstall.packages(c("ape","dplyr", "magrittr", "igraph", "viridisLite", "MASS", "shiny"))\n<\/pre><\/div>\n\n\napp.R<\/code>. This app.R<\/code> allows you to build a shiny<\/a> application.<\/p>\n\n\n\nrunApp()<\/code> inside the console and your browser will open the app.<\/p>\n\n\n\nsideBarPanel<\/code> that has all the inputs we need<\/p>\n\n\n\nserver<\/code> function that will provide<\/p>\n\n\n\nbiopsy<\/code> and biopsy_numeric()<\/code>, as biopsy_numeric<\/code> is a reactive<\/a>.<\/p>\n\n\n\nThe input data set<\/h3>\n\n\n\n
biopsy<\/code> inside the app. The columns 2-10 were stored inside the reactive biopsy_numeric()<\/code> which is filtered by the input$patients<\/code> input to not use all 699 patients, but between 1 and 100.<\/p>\n\n\n\n1) Construction of a
SelectInput<\/code><\/h3>\n\n\n\nSelectInput<\/code> shall allow the user to not use all 9 measured variables, but just the ones he desires. This shall help finding the measurement, that is necessary to classify patients. What is a shiny selectInput? We can therefore look at the description of the selectInput by<\/p>\n\n\n\n?shiny::selectInput\n<\/pre><\/div>\n\n\n
![\"\"\/](\"https:\/\/cdn-images-1.medium.com\/max\/800\/0*jH0mGpgdpTmNSSXA.jpg\")
<\/figure>\n\n\n\nchoices<\/code> as the column names of the biopsy data set from 2-10. the selected<\/code> input will be the same. We shall allow multiple inputs, so multiple<\/code> will be set to TRUE<\/code>. Additionally we shall name the inputId<\/code> “vars”. So we can replace the part output$variables<\/code> inside the app.R<\/code> file with this:<\/p>\n\n\n\noutput$variables <- renderUI({\n selectInput(inputId="vars",\n label = "Variables to use",\n choices = names(biopsy)[2:10],\n multiple = TRUE,\n selected = names(biopsy)[2:10]\n )\n})\n<\/pre><\/div>\n\n\n2) A Heatmap to see the outcome of clustering<\/h3>\n\n\n\n
![\"\"\/](\"https:\/\/cdn-images-1.medium.com\/max\/800\/0*hFbSaqo3RErXjxUr.jpg\")
<\/figure>\n\n\n\n