Fig. 5.

Validation of novel blood RNA classifiers as a diagnostic for Alzheimer’s disease. We used the independent batch 2 AD data set (see “Materials and methods”) to test the predictive performance of our healthy ageing classifier and our previously published AD prototype diagnostic. The performance of each was evaluated using ROC curves. The healthy ageing gene classifier generated independent AUCs of 0.73 and 0.66 for AD in cohorts 1 and 2, respectively. For the combined ‘healthy ageing’ plus ‘AD disease’ RNA classifier (150 + 48 probe-sets) we obtained AUCs of 0.86 and 0.73 for AD without any attempt at optimization. The AD disease RNA classifier probe-sets were selected using cohort 1