22-zero-non-diseased-cases.Rmd

# Analyzing a dataset with only diseased cases {#analyze-diseased-only-dataset}

---
output:
  rmarkdown::pdf_document:
    fig_caption: yes        
---

```{r setup, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>",
  eval.after = "fig.cap"
)

library(RJafroc)
```




## TBA How much finished {#analyze-diseased-only-dataset-how-much-finished}
0%


## The problem {#analyze-diseased-only-dataset-methods}

How to analyze $K_1 = 0$ datasets.

ROC-like plot of TPF vs. FPF1 is possible, see Section \@ref(empirical-definition-empirical-auc-afroc1). Can create a ROC-like dataset with equal number of "non-diseased" and diseased cases (the ratings of the non-diseased cases are the FP ratings on diseased cases). Fit RSM to this dataset. Proceed as before. Key assumption being violated: the FP ratings on diseased cases are independent of the TP ratings on same cases. However, without this assumption one cannot estimate RSM parameters. Need `RJafroc` function to handle this special case: `FitRsmRoc1`? No! Just need function to create a "ROC" dataset from one that only has diseased cases. e.g., `DfNoNormalsDataset`?


### Step 1: Create a test (diseased cases only) dataset {#analyze-diseased-only-dataset-step-1}

Save TONY dataset to `dsTony`. Create copy `dsNoNormals`. Remove all normal cases from it. 

```{r analyze-diseased-only-code, echo=TRUE}
dsTony <- RJafroc::dataset01 # TONY dataset
K2 <- length(dsTony$lesions$perCase)
K1 <- length(dsTony$ratings$NL[1,1,,1]) - K2
dsNoNormals <- dsTony
# Remove all normal cases
dsNoNormals$ratings$NL <- dsNoNormals$ratings$NL[,,-(1:K1),] 
# And fix truthTableStr
dsNoNormals$descriptions$truthTableStr <- 
  dsNoNormals$descriptions$truthTableStr[,,-(1:K1),]
RJafroc::UtilFigureOfMerit(dsTony,FOM = "wAFROC")
#RJafroc::UtilFigureOfMerit(dsNoNormals,FOM = "wAFROC") #this will generate an error
RJafroc::UtilFigureOfMerit(dsTony,FOM = "wAFROC1")
RJafroc::UtilFigureOfMerit(dsNoNormals,FOM = "wAFROC1")
st <- St(dsTony,FOM = "wAFROC")
st1 <- St(dsNoNormals,FOM = "wAFROC1")
st$RRRC
st1$RRRC
```


* `dsNoNormals` is the dataset with no non-diseased cases. 

* `st` contains the results of significance testing using the wAFROC-AUC figure of merit for the full dataset. 

* `st1` contains the results of significance testing using the wAFROC1-AUC figure of merit for the dataset with no non-diseased cases.