JMIR Publications

ABSTRACT

Background:

Discriminating mild cognitive impairment (MCI) from subjective cognitive decline (SCD) and healthy controls (HC) remains challenging. Digital tools can overcome classic paper-and-pencil test limitations, offering sensitive and engaging screening for early cognitive impairment.

Objective:

To evaluate the diagnostic accuracy of CogniPaz, an ecologically oriented, tablet-based, gamified cognitive assessment tool composed of nine tasks designed to evaluate memory and executive functions, in discriminating MCI from SCD and HC.

Methods:

This prospective observational study recruited participants with cognitive complaints and HC at La Paz University Hospital Cognitive Impairment Unit (April 2025–January 2026). Following clinical and Montreal Cognitive Assessment (MoCA) evaluations, participants were classified into HC (no cognitive complaints, MoCA≥26), SCD (cognitive complaints, MoCA≥26), and MCI (cognitive complaints without functional impairment, MoCA<26; subdivided into amnestic [aMCI] and non-amnestic phenotypes). An evaluator blinded to MoCA results oversaw the automated, self-administered CogniPaz assessment (including automatic timing). Receiver operating characteristic (ROC) analyses (with age-adjusted models compared via DeLong's test) evaluated diagnostic accuracy, Spearman assessed CogniPaz-MoCA correlation, and a 0-9 scored questionnaire measured CogniPaz usability. A subset of participants with available biomarker data underwent an exploratory analysis to assess the tool's capacity to distinguish biologically confirmed neurodegeneration suggestive of Alzheimer Disease (AD).

Results:

Participants (N=84; median age: 72.0 years) included HC (n=27; 74.0), SCD (n=17; 59.0), and MCI (n=40; 74.0) groups, the latter subclassified into aMCI (n=22; 76.5) and non-amnestic MCI (n=18; 71.5). The aMCI group was significantly older than the SCD group (P=.003). The median administration time for CogniPaz was 9.39 minutes (IQR 7.53–11.38). CogniPaz scores correlated strongly with MoCA (r=.81, 95% CI .75-.89, P<.001). CogniPaz achieved high diagnostic accuracy, independently of age: MCI vs HC (AUC=0.880, 95% CI 0.798-0.961), MCI vs SCD (AUC=0.915, 95% CI 0.841-0.990), aMCI vs HC (AUC=0.910, 95% CI 0.830-0.990), and aMCI vs SCD (AUC=0.943, 95% CI 0.878-1.000). It showed capacity for detecting biologically confirmed neurodegeneration (n=39, AUC=0.810, 95% CI 0.661-0.958). At the<28.5 points cutoff, sensitivity ranged from 85.0% to 90.9% and specificity from 74.1% to 82.4% across comparisons. Particularly, it was accurate in differentiating aMCI vs SCD (90.9% sensitivity, 82.4% specificity). User satisfaction was high (median 8.4/9).

Conclusions:

CogniPaz showed good capacity to differentiate MCI/aMCI from SCD and HC, and appears promising for identifying AD-related neurodegeneration, with a strong correlation with MoCA score and good patient acceptance. These results support its use as a digital screening tool for early-stage cognitive impairment. External, multi-centre longitudinal validation is essential before considering its implementation in primary care settings.