Gender imbalance in AI training datasets lowers results

Artificial intelligence (AI) algorithms trained with more cases from one gender...Read more on AuntMinnie.comRelated Reading: AI can differentiate normal, abnormal chest x-rays AMVC 2020: What now for artificial intelligence? AMVC 2020: Artificial intelligence for COVID-19 detection Free AI software can help triage COVID-19 on x-ray AI may predict COVID-19 progression on chest x-raysComments: 6/9/2020 1:16:42 PMDXAman From the article: "This raises the alarm for national agencies in charge of regulating and approving computer-assisted diagnosis systems, which should include explicit gender balance and diversity recommendations," the authors wrote. "We also establish an open problem for the academic medical image computing community which needs to be addressed by novel algorithms endowed with robustness to gender imbalance."   How about a separate male and female training set and a separate model for each sex, rather than trying to combine them because being inclusive is more important than being accurate?  They should not then be surprised when a CXR model trained on mostly female CXR's fails when looking at male lungs that lack breast shadows?  Or the reverse?  Did they really need to do this study?
Source: AuntMinnie.com Headlines - Category: Radiology Source Type: news