03-12-2017 07:41 PM
Hello, I am working on using ROC curves using the Proc Logistic. I have used my ROC curves for what I need but am trying to create a table for 4 different models that shows the Sensitivity/Specificity/PPV/and NPV that looks similar to the attached Untitled JPEG. I am trying to use a cutpoint of 95% sensitivity, so i was able to pull up the dataset table created by the proc logistic (see the excel output) and I went down to the sensitivity of 95% and found a 1-specifcity of .23. How do i go about getting confidence intervals for the sensitviity, specificty and how do i calculate the ppv/npv as well from this? I am looking at a score that has continuous variables from 1-24 to see its accuracy in prediciting an outcome of alive or dead.
Should I be doing this a different way than proc logistic? My code is:
proc logistic data= test.n; model dead (ref = first) = rems_new /outroc =roc1; roc; run;
07-03-2017 01:53 PM
If you pick a cutpoint on the predicted probabilities (such as the cutpoint that gave the sensitivity you mentioned), then you can use the value of that cutpoint to classify the observations as predicted events or nonevents. You can then use PROC FREQ to show the table of predicted by actual events and nonevents and get the statistics you want, along with confidence intervals, as shown in this note.