MemTrax as a scalable digital triage tool for mild cognitive impairment screening: primary validation in Alzheimer’s disease and exploratory analysis in Parkinson’s disease

Abstract

Alzheimer’s disease (AD) and Parkinson’s disease (PD) pose escalating global health challenges, with early detection of mild cognitive impairment (MCI) being critical for timely intervention. The 2023 Delphi consensus highlights the need for scalable tools suitable for primary care. This study aimed to validate the MemTrax as a rapid memory-focused triage tool against consensus-recommended cognitive assessments, including the Mini-Mental State Examination (MMSE) and the Montreal Cognitive Assessment (MoCA), for detecting MCI due to Alzheimer’s disease (AD-MCI) as the primary objective, and to explore its performance in MCI due to Parkinson’s disease (PD-MCI) as an exploratory aim, evaluating its utility as a rapid, accessible digital tool. We conducted a cross-sectional study enrolling 222 participants: 74 with AD-MCI, 73 healthy controls (HCs; primary cohort); 42 PD patients with normal cognition (PD-NC), 33 with PD-MCI (exploratory cohort). All participants underwent MMSE, MoCA and MemTrax testing. Diagnostic accuracy, optimal cut‑offs, and correlations between MemTrax parameters (percent correct [MTx‑%C], reaction time [MTx‑RT], composite score [MTx‑Cp]) and standard cognitive measures were evaluated. In the primary AD cohort, MTx-%C showed an area under the curve (AUC) of 0.876 (95% CI: 0.813–0.926) for discriminating AD-MCI from HCs, comparable to MoCA (AUC = 0.879) and MMSE (AUC = 0.916). At a cut-off of ≤ 80%, MTx-%C achieved a sensitivity of 70.3% and specificity of 90.4%. In the exploratory PD cohort, MTx-%C (AUC = 0.808) showed similar discrimination to MMSE (AUC = 0.694) in differentiating PD-MCI from PD-NC (DeLong’s test, p = 0.142). Multivariate analysis of MTx-%C, MTx-RT, and MTx-Cp demonstrated significant separation between MCI and cognitively normal groups in both the AD and PD cohorts. MTx-%C showed significant correlations with MMSE in AD ( r = 0.709) and PD ( r = 0.340), and with MoCA in AD ( r = 0.601) and PD ( r = 0.608). MemTrax metrics correlated significantly with distinct domain-level associations reflecting disease-specific cognitive profiles. MemTrax demonstrates good diagnostic accuracy comparable to established cognitive screening tools for AD-MCI (primary cohort), supporting its use as a rapid, scalable, and culturally neutral digital memory-focused triage tool. Exploratory findings in PD-MCI are promising but require confirmatory studies. Its brevity and automation support practical implementation in primary care and community settings, potentially facilitating earlier detection and stratification of cognitive impairment.