Mood and Age Predict Cognitive Complaints in Memory Clinic Patients: A Machine-Learning and Linear Modeling Approach.

Introduction Cognitive complaints are often considered early indicators of Alzheimer's disease (AD) and commonly lead to memory clinic consultations. Prior studies suggest stronger associations between cognitive complaints and mood than with objective cognition, but this interplay remains poorly understood. Using a machine learning-supported approach, we aimed to (1) identify key predictors of cognitive complaints, and (2) compare the value of gamified versus standard neuropsychological testing in detecting subtle deficits. Methods In this international multi-center study, 98 participants (57 females; mean age 71.9, range 55-86) from three memory clinics completed the Cognitive Failures Questionnaire (CFQ), mood and apathy questionnaires, the tablet-based gamified Adaptive Cognitive Evaluation Explorer (ACE-X), and standard neuropsychological tests. Predictors of CFQ scores were examined using elastic net regression and the Boruta algorithm, followed by linear mixed-effects modeling. Results Greater mood symptoms were associated with more cognitive complaints, whereas increasing age was linked to fewer complaints. Study center accounted for additional variance. The final model explained a substantial proportion of variance (conditional R2 = 0.48, marginal R2 = 0.33). Participants had lower z-scores on ACE-X compared to standard testing, but neither predicted the severity of cognitive complaints. Discussion Mood and age were main predictors of cognitive complaints in memory clinic patients. Although ACE-X yielded lower normative scores than standard tests, neither cognitive measure was linked to complaints. These findings highlight the importance of systematically assessing mood, adopting personalized approaches when evaluating subjective and objective cognition, and the potential value of gamified assessments for screening populations at risk of AD.