Shiny plotOutput with plot_features from the lime package produces nothing

[ylu73]

Hi I am a beginner user of lime and shiny and am attempting a project to create a shiny app. I want to develop a risk calculator using a random forest model I trained in R ("rffit.rda"). similar to this web app calculator

https://sorg-apps.shinyapps.io/thaopioid/

But the prediction panel in my app gives me no output. I was able to get the code to execute outside of shiny in the regular R environment, but when I add the prediction and explanation functions to the server end nothing appears when I run the app. Grateful for any help.

library(shinydashboard)
library(lime)
library(caret)
library(dplyr)
load("rffit.rda")
ui <- dashboardPage(
  dashboardHeader(title = "Postoperative Opioid Consumption Risk Calculator", 
                  titleWidth = 500),
  dashboardSidebar(
    sidebarMenu(
      menuItem("Welcome", tabName = "welcome", icon = icon("dashboard")),
      menuItem("Input", tabName = "input", icon = icon("th")),
      menuItem("Prediction", tabName = "predictions", icon= icon("th"))
    )
  ),
  dashboardBody(
    tabItems(
      # First tab content
      tabItem(tabName = "welcome",
              h2("Disclaimer"),
              h3(strong("This tool is designed for general educational purposes only and is not intended in any way to substitute
                    for professional medical advice, consultation, diagnosis, or treatment. Any analysis, report, or information
                    contained in or produced by this tool is intended to serve as a supplement to, and not a substitute for the knowledge, 
                    expertise, skill and judgment of health care professionals. In no event shall this tool under this Agreement, 
                    be considered to be in any form, medical care, treatment, or therapy for patients or users of this tool.")),
              h3("This tool's services are provided 'as is'. These services provide no warranties, express or implied and shall not be
             liable for any direct, consequential, lost profits, or other damages incurred by the user of this information tool.")
            ),
      # Second tab content
      tabItem(tabName = "input",
              selectInput("preop_narc", "Opioid use during the preoperative period (1 year to 30 days before surgery); 1=Yes, 0=No", 
                          choices = c("1", "0"), selected = "Yes"),
              numericInput("periop_ome", "Total morphine equivalent consumed during the perioperative period (30 days before surgery to 15 days after)", min = 0, value = 0),
              numericInput("unemployment", "Community percent unemployment", min = 0, value = 0),
              numericInput("med_inc", "Median household income($)", min = 0, value = 0),
              numericInput("hs", "Community percent high school graduate or GED obtained", min = 0, value = 0),
              numericInput("poverty", "Community percent living at poverty line", min = 0, value = 0),
              sliderInput("age", "Age", 0, 120, 0),
              sliderInput("preop_pain", "Preoperative pain", 0, 10, 0),
              numericInput("days_symptoms", "Days from symptom onset to surgery", min = 0, value = 0),
              actionButton("goButton", "Go!")
      ),
      # Third tab content
      tabItem(tabName = "predictions",
              plotOutput("explanations")
    )
  )
)
)
server <- function(input, output) {
  predictions <- eventReactive(input$goButton, {
  req(input$preop_narc, input$periop_ome, input$unemployement, input$med_inc, input$hs, input$poverty, input$age, input$preop_pain, input$days_symptoms)
  inputdata <- cbind(input$preop_narc, input$periop_ome, input$unemployement, input$med_inc, input$hs, input$poverty, input$age, input$preop_pain, input$days_symptoms)
  colnames(inputdata) <- c("narc", "preop_total_ome_1",
  "Percent__EMPLOYMENT_STATUS___Population_16_years_and_over___In_labor_force___Civilian_labor_force___Unemployed",
  "medinc", "Percent__Estimate__Percent_high_school_graduate_or_higher", "pov_100", "age_1", "Rate_your_pain_on_a_scale_from_1_10__1__minimal_pain__10__severe_pain__", "symptom_duration")
  inputdata$narc <-as.factor(inputdata$narc)
  training_set <- read.csv("training_set.csv")
  final_data <- rbind(training_set, inputdata)
  prediction = caret::predict(rffit, final_data, type = "raw")
  outputdata = cbind(final_data, prediction)
  outputdata
})

output$explanations <- renderPlot({
    pred = predictions()
    pred_1 <- lime(pred, rffit, bin_continuous = TRUE, quantile_bins = FALSE)
    pred_2 <- lime::explain(pred[1205,], pred_1, n_labels = 1, n_features = 9)
    pred_2$feature_desc <- c("Preoperative Opioid Use", 
                             "Perioperative 1 Year Opioid Consumption (OME)", 
                             "Percent unemployment", 
                             "Median income", 
                             "Percent high school graduate", 
                             "Percent living at poverty line", 
                             "Age", 
                             "Preoperative pain", 
                             "Duration of symptoms < 2Y")
    explain_plot <- plot_features(pred_2, ncol =1)
    explain_plot
})
}
shinyApp(ui, server)