B - 90 Establishing Cutoffs for Impaired Cognition in Virtual Reality Assessment: the Virtual Kitchen Protocol for Learning and Memory

Abstract Objective The Virtual Kitchen Protocol for Learning and Memory (VKP-LM) is a virtual-reality-based measure of everyday functioning through procedural memory tasks. Previous literature has introduced the VKP-LM and begun validation processes. This study explores cutoff scores to determine cognitive impairment across the Immediate (VKP-IR) and Delayed-Recall (VKP-DR) portions of the VKP-LM. Method Adults (N = 126; Mage = 70.29; 56.3% Female) completed the VKP-LM as a part of a neuropsychological evaluation in a community-based assessment center. Participants were categorized binarily as to whether they received a neurocognitive disorder (NCD) diagnosis. Discriminatory abilities and cutoffs were determined by the Receiver Operator Curve’s (ROC) Area Under the Curve (AUC) and sensitivity-specificity charts. Classification tables determined the Positive Predictive Power (PPP) and Negative Predictive Power (NPP) given the established cutoffs. Results ROC analysis showed the VKP-IR (Max = 70; Cutoff = 28) to have good discriminatory abilities (AUC = 0.788) and the VKP-DR (Max = 70; Cutoff = 26;) to have excellent discriminatory abilities (AUC = 0.855). The classification tables showed that the VKP-IR had a PPP = 63.0% and NPP = 87.9%, while the VKP-DR had a PPP = 86.4% and NPP = 90.4%. Conclusion Due to the multifaceted nature of diagnosing NCDs, single-domain measures (i.e., procedural memory) will always struggle to obtain excellent PPP, as not all NCDs have amnestic cognitive profiles (e.g., cerebral vascular, Lewy body disease, etc.). Further research should explore comparing discriminatory abilities against other measures, increasing the sample size (both groups), and comparing discriminatory abilities among only NCDs with amnestic cognitive profiles, such as Alzheimer’s disease.