How to calculate specificity, sensibility, predictive values and ROC curve in R?-CodePudding

I have a rather small dataset resulted from a linkage between two different datasets. I would like to know how can I calculate specificity, sensibility, predictive values and plot the ROC curve. This is the first time I'm using this kind of statistics in R, so I don't even know how to start.

Part of the data looks like this:

data <- data.frame(NMM_TOTAL = c(1, 1, 0, 1, 1, 1, 1, 0, 0, 1, 1, 0, 0, 1, 1),
                   CPAV_TOTAL = c(0, 1, 1, 0, 1, 1, 0, 0, 0, 0, 1, 1, 1, 0, 0),
                   SIH_NMM_TOTAL = c(0, 0, 0, 1, 1, 1, 1, 1, 1, 0 , 0, 1, 1, 0, 1),
                   SIH_CPAV_TOTAL = c(1, 1, 0, 1, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1))

And the two way tables would be the combination of:

tab1 <- table(data$SIH_NMM_TOTAL, data$NMM_TOTAL)

tab2 <- table(data$SIH_CPAV_TOTAL, data$CPAV_TOTAL)

Where NMM_TOTAL and CPAV_TOTAL are the "gold standard". I don't know if any of this makes sense. Thanks in advance!

Obs: 1 stands for positive and 0 for negative.

CodePudding user response：

Let's work with tab1 to demonstrate specificity, sensitivity, and predictive values. Consider labeling the rows and columns of your tables to enhance clarity

  act <- data$SIH_NMM_TOTAL
  ref <- data$NMM_TOTAL
  table(act,ref)

Load this library

library(caret)

The input data needs to be factors

  act <- factor(act)
  ref <- factor(ref)

The commands look like this

  sensitivity(act, ref)
  specificity(act, ref)
  posPredValue(act, ref)
  negPredValue(act, ref)

ROC curve. The Receiver Operating Characteristic (ROC) curve is used to assess the accuracy of a continuous measurement for predicting a binary outcome. It is not clear from your data that you can plot an ROC curve. Let me show you a simple example on how to generate one. The example is drawn from https://cran.r-project.org/web/packages/plotROC/vignettes/examples.html

library(ggplot2)
library(plotROC)

set.seed(1)
D.ex <- rbinom(200, size = 1, prob = .5)
M1 <- rnorm(200, mean = D.ex, sd = .65)

test <- data.frame(D = D.ex, D.str = c("Healthy", "Ill")[D.ex   1], 
                   M1 = M1, stringsAsFactors = FALSE)
head(test)

ggplot(test, aes(d = D, m = M1))   geom_roc()