If this gets cheap enough and the predictive values hold up in ordinary clinical populations, it could change when people actually get evaluated rather than just how they get evaluated