would have a higher predictor score than an individual without the disease, and therefore reflects the proportion of individuals classified correctly as cases or controls. An AUC of 0.5 means that the predictor can accurately classify 50% of individuals, or no greater than chance, whereas an AUC of 1.0 means that the predictor can correctly classify 100% of individuals. An AUC of 0.80 is generally accepted as a target cut-off for screening and 0.99 for diagnosis (Janssens et al., 2006). Simulation studies that we have conducted suggest that if all genetic contributions are included in a prediction model for AD, given AD’s heritability of around 50%, there is the potential for AUCs approaching 0.80 to be reached with genetic information alone (Maher et al., in preparation).
