Fig. 4.Receiveroperatingcharacteristic(ROC) curvesofELISAdatafromthe‘validation’cohort.ROCanalysesof the‘validation’cohortELISAdatawereperformedforeachbiomarkerto distinguishgoodcognitionfrompoorcognition.Astep-wiselogisticregressionmodelwasappliedtoidentifyacomplementarycombinationofbiomarkersthatwouldoptimizeaccuracy (maximize area under the curve (AUC) without including additional non-contributorybiomarkers,accepted). Preoperative (A) CSF p-tau, AUC = 0.733, p = 0.046; (B) Aß1–38/Aß1–42, AUC = 0.804, p = 0.009, and(C) Aß1–42/p-tau's AUC = 0.753, p = 0.017.