Predictive utility of a priori algorithmic definitions of prAD for progression
to dementia 
   Lead Investigator:    Heather Romero
   Institution      :    Duke University Medical Center
   E-Mail           :    heather.romero@duke.edu
   Proposal ID      :    791



   Proposal Description:
   The analytic approach employs a priori algorithmic definitions of prAD to determine its predictive utility for progression to dementia in the NACC database. This strategy assembles individuals into groupings based on cut scores on quantitative measures of episodic memory, executive function, and functional ability, a method we have used in previous studies. Using published normative standards for each test, we will transform neuropsychological scores to uniform metrics of normally disturbed z scores using the formula (raw score - normative mean)/SD for each test). Accuracy for identifying incident AD cases will be determined by calculating sensitivity, specificity, positive predictive values, and reversion to normal cognition over a follow-up observational period for all five transitional states. These values will be compared in order to determine diagnostic stability of prAD relative to other criteria for MCI. Sensitivity and specificity of each transitional state will be compared using Receiver Operating Characteristic (ROC) curves to evaluate the accuracy of prAD relative to other MCI criteria. The gold standard for correct classification will be based on clinical diagnosis made by NACC investigators.