Cognitive modeling of the Mnemonic Similarity Task as a digital biomarker for Alzheimer's Disease

Abstract

AD related pathologies, such as beta-amyloid (A{beta}) and phosphorylated tau (pTau), are evident decades before any noticeable decline in memory occurs. Identifying individuals during this asymptomatic phase is crucial for timely intervention. The Mnemonic Similarity Task (MST), a modified recognition memory task, is especially relevant for early AD screening, as it assesses hippocampal integrity, a region affected (both directly and indirectly) early in the progression of the disease. Further, strong inferences on the underlying cognitive mechanisms that support performance on this task can be made using Bayesian cognitive modeling. We assessed whether analyzing MST performance using a cognitive model could detect subtle changes in cognitive function and AD biomarker status prior to overt cognitive decline. We analyzed MST data from >200 individuals (young, cognitively healthy older adults, and individuals with MCI), a subset of which also had existing CSF A{beta} and pTau data. Traditional performance scores and cognitive modeling using multinomial processing trees was applied to each participants MST data using Bayesian approaches. We assessed how well each could predict age group, memory ability, MCI status, A{beta}/pTau status using ROC analyses. Both approaches predicted age group membership equally, but cognitive modeling approaches exceeded traditional metrics in all other comparisons. This work establishes that cognitive modeling of the MST can detect individuals with AD prior to cognitive decline, making it a potentially useful tool for both screening and monitoring older adults during the asymptomatic phase of AD.