« Biological sex classification with structural MRI data shows increased misclassification in transgender women »
I tried to check a study about those brain scans stuff. Their conclusion is not « trans brain stuff is bogus » but when you look at the data…
The Boxplot shows that the classifier still tends to correctly assign the sex even in transgender women. The vast majority of trans women in their data set were post-hormone treatment. And finally, when you check supplementary material S3, their classifier is correct 99% of the time on cisgender women and 88% of the time on men. Given that they only had trans women amongst their trans, I suspect they have somehow chosen a biased model in the first place, in order to have more trans women be classified as women. It seems biased just enough so that the problem does not appear on the Boxplot of cisgender men.
« Biological sex classification with structural MRI data shows increased misclassification in transgender women »
I tried to check a study about those brain scans stuff. Their conclusion is not « trans brain stuff is bogus » but when you look at the data…
The Boxplot shows that the classifier still tends to correctly assign the sex even in transgender women. The vast majority of trans women in their data set were post-hormone treatment. And finally, when you check supplementary material S3, their classifier is correct 99% of the time on cisgender women and 88% of the time on men. Given that they only had trans women amongst their trans, I suspect they have somehow chosen a biased model in the first place, in order to have more trans women be classified as women. It seems biased just enough so that the problem does not appear on the Boxplot of cisgender men.
Nature is such a fraudulent journal in 2024.