We thank Dr Dieck and co-authors
1 for the comments on our paper.
2 They performed an important experiment. In that, ophthalmologists were told about some features of the difference in the fundus image of men and women in advance and judged the sex of each eye from her/his fundus photograph. As a result, the correct judgment rate was only around 60%. The first author of our paper, Takehiro Yamashita, also did judgment using the same software and the accuracy rate was about the same (personal communication with Dr Nikolas Pontikos). On the other hand, Poplin et al. reported that deep learning artificial intelligence (AI) gave a far better judgment rate of 97%.
3 In the statistics, it has been considered important to eliminate elements that have a small effect in order to avoid statistical errors including multicollinearity (e.g. stepwise analysis). However, in terms of discrimination, it can be said that inserting multiple elements is superior even if one element alone has no significant effect. Thus, comprehensive judgment using many factors by AI may be getting more useful and popular for predicting the results.
Most importantly, what the present study means would not be just the development of AI algorism discriminating between sexes. We believe this means the appearance of completely new way of research. Namely, AI came up with a theme that humans had never thought of, and then humans find out the truth by analyzing that theme. When this method spreads in the future, it will be discovered that previously unanticipated factors are important for the certain phenomena more easily and rapidly. These studies show that AI can be a new teacher for humans in research.