The Digital Maze Test (dMaze) Reveals Subtle Decision‐Making Difficulties in MCI: Hurrying to Keep Up?

Abstract

AbstractBackgroundTraditional Maze tests are time‐consuming and insensitive to early AD cognitive changes. We combine behavior capture using a digitizing pen with novel test design to produce person‐centric/relative metrics that disembed cognitive from motor speeds. This enables comparison of complex decision‐making behavior across difficulty levels within a single test. We compare traditional measures with our person‐centric/relative measures.MethodParticipants were healthy (HC = 22), Mild Cognitive Impairment (MCI = 37), Alzheimer’s (AD = 39), and Parkinson’s (PD = 43) volunteers (N = 141). Diagnosis was determined by consensus using standard of care data (neuropsychological/neurological evaluations, MRI and DAT scans). Participant groups differed in age: HC (52.27) younger than PD (64.07), MCI (75.05) and AD (75.38) oldest. Groups differed in the expected direction for DCTclock and MoCA scores.The dMaze test contains a path‐following condition without choice points, followed by a choice (CH) condition with identical solution paths/motor demands. CH contains simple and complex choices. Relative measures use the participant’s overall speed to normalize within‐test speed deviations. We examined: Total Time (TTime) to completion, Max motor speed (MaxSpd) and Median speed (MedSpd), Choice point Average Speed (AveSpd) and Relative Average Speed (rAveSpd).ResultTTime in seconds differentiated (p = .001) AD (141.24) from HC (52.98) and PD (77.35), not MCI (103.16). MaxSpd was slowest (p = .004) for HC (59.82mm/sec) compared to all clinical groups (MCI = 76.71mm/sec, PD = 77.11, AD = 85.52). Groups did not differ for MedSpd or choice point AveSpd. Simple choice point rAveSpd differentiated only AD (p = .001) from HC, PD & MCI. Complex choice point rAveSpd differentiated HC from MCI (p = .05) and AD (p = .001). HC and PD did not differ, suggesting simple motor speed did not account for ink speed differences.ConclusionFast completion time despite slow maximum speed and relative average speed across choice points suggests HC use a modulated approach balancing speed and task difficulty. MCI faster relative maximum speed (hurrying) with completion time similar to HC, suggests a less well‐modulated approach. Relative speed differences for more complex choices may be sensitive to preclinical cognitive change even when total completion time is not. The ability to precisely measure and parse person‐centric motor and cognitive test behaviors provides opportunity to improve preclinical detection, disease monitoring and treatment efficacy.