Youden's index

Question

[js1234]

Hello everyone,

I want to analyze the expression of a microRNA and its effect on survival of cancer patients.

For that i want to divide the samples into two groups according to their expression levels (low expression of miR and high expression of miR). To establish the cut-off point for this division can i use the Youden's index when log-rank test p-value is > 0.05 or the p value of AUC is > 0.05?

Or when this occurs should i use median of the expression data as cut-point to establish the two groups?

Thanks

[Hamid Ghaedi]

I would say you could use z-score normalization or maximally selected rank statistics to determine up/down regulated cases with regards to gene expression.