Chekroud and colleagues provided evidence that an individual's biological sex can be classified with high accuracy from the brain's "mosaic" patterns. They were able to accurately predict sex with an accuracy of 93% in a large held-out sample. (4/8)https://doi.org/10.1073/pnas.1523888113 …
-
Show this thread
-
How well can we establish sex from the differences in neuroanatomical features? A team of researchers found that they could accurately predict sex with 83% accuracy in cross-validated set and 77% in independent data set. (5/8) https://doi.org/10.1016/j.neuroimage.2018.01.065 …
1 reply 1 retweet 9 likesShow this thread -
How about patterns in the gray matter? A team finds that "Models using components of brain gray matter volume and concentration were able to differentiate between males and females with greater than 93% generalizable accuracy." (6/8)https://doi.org/10.1002/hbm.24462 …
2 replies 3 retweets 12 likesShow this thread -
Patterns in the functional connectivity of the brain is related to sex and ancestry. A team provides evidence that ancestry and sex can be quite accurately predicted from these patterns. Area-under-curve of 0.98 for prediction of sex. (7/8)https://doi.org/10.1101/440776
1 reply 2 retweets 14 likesShow this thread -
While none of these predictions have perfect accuracy, it is clear that we can do much better than chance. How well might we be able to do if all of the information above was used simultaneously? We don't know for sure, but it would be a very high accuracy. Thread over. (8/8)
3 replies 2 retweets 21 likesShow this thread -
Replying to @Scientific_Bird
Features: N, accuracy, and model: brain waves: 1308, 80%, CNN functional connectivity: 820, 87%, regression brain 'mosaic': 1566, 93%, regression neuroanatomy: 967, 77%, SVM grey matter: 1300, 93%, most shallow models including SVM functional connectivity: 950, 98%, elastic net
1 reply 0 retweets 2 likes -
Replying to @AndrewCutler13 @Scientific_Bird
From a computer vision standpoint, these are naive models trained on very little data. Check out performance boosts in the ImageNet classification competition, mainly driven by larger models that can leverage info from very large N:pic.twitter.com/ibSjB8CGUg
2 replies 0 retweets 2 likes -
Replying to @AndrewCutler13 @Scientific_Bird
From a modeling and N standpoint, these studies which yield 80-90% acc are not even at the 2010 point. My guess is if there was a good dataset and a competition, gender classification would go to ~100% acc.
1 reply 0 retweets 1 like -
Replying to @AndrewCutler13 @Scientific_Bird
Andrew Cutler Retweeted Emil O W Kirkegaard
Relevant, there seems to be a publication bias for reporting smaller gender differences (reversing the natural tendency to oversell):https://twitter.com/KirkegaardEmil/status/1018494702977474561 …
Andrew Cutler added,
0 replies 0 retweets 1 like
There is a meta on brain size differences. You should check it for pub bias.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.