# What does k mean in terms of sensitivity and specificity

I am learning a paper Assessment of Deep Generative Models for High-Resolution Synthetic Retinal Image Generation of Age-Related Macular Degeneration, which uses an item "K" in the stats of sensitivity and specificity.

I googled a bit and didn't get a related hit. Could someone please give a hint? Thanks in advance.

That's not a 'k', its the letter Kappa, probably Cohen's Kappa, describing the statistic which typically measures the inter-rater agreement, but is sometimes also used for intra-rater agreement. That seems to be the case in this paper.

• +1, because that is probably what it is supposed to be. However, I am puzzled. $\kappa$, as you write, measures the agreement between two raters in a multi-category classification task. But here, the paper reports separate $\kappa$s for each rater. This does not seem to make sense. The OP's best bet may be to contact the corresponding authors. – Stephan Kolassa Apr 7 '20 at 8:14
• Good point about Kappa typically referring to inter-rater agreement! I briefly looked into the wiki article, and apparently Cohen's Kappa can also be used for intra-rater agreement, which is probably the case here. Edited the answer accordingly – S. Warmelink Apr 7 '20 at 8:19