Generative adversarial networks to predict treatment response for neovascular age-related macular degeneration: interesting, but is it useful?