For example, I have a dataset of HE4 in the form of a numerical variable. and I want to apply it to a survival analysis for ovarian cancer prognosis. How can I find a cut off of HE4 value which is the best for prognosis of survival outcome?
I faced a similar problem some time ago. As far as I know, one approach you can use is Maximally Selected Rank Statistics, which can select the optimal cutoff point of a continuous variable to maximise the predictive value of a survival analysis. I performed such analysis using R with the "maxstat" package, you can find more information here:
I think it depends firstly on the type of data you are analyzing (for example, transcriptomics data, immunohistochemistry data, etc.) and if there are any previously reported (and defined) cutoff values (this might be the case for immunohistochemistry scores for example).
In any case, I recommend to use the Cutoff finder tool ( http://molpath.charite.de/cutoff/index.jsp ), as Peter Nagy suggested. Once in the website, you'll be able to choose among a couple of methods to determine a cutoff value (one of these methods is based on survival analysis).