Refining the clinical interpretation of activity variability in cognitive impairment: The need for phenotypic specificity

Abstract

To the Editor,

We read with great interest the recent article by Donahue et al. entitled “Activity variability: a novel physical activity metric and its association with cognitive impairment.”¹ The authors proposed an innovative metric based on minute-to-minute accelerometry data to quantify behavioral complexity in older adults and demonstrated its strong association with cognitive impairment. This approach represents a valuable step forward in moving beyond threshold-based activity summaries, such as daily activity counts or activity fragmentation. The findings suggest that reduced activity variability may serve as a potential behavioral biomarker of cognitive decline, with promising implications for early detection.

From the perspective of neurological clinical care, however, we believe there is an opportunity to further refine the interpretation and application of activity variability by considering the heterogeneity of cognitive impairment phenotypes. While cognitive decline is often grouped into a single binary or trichotomous classification (e.g., no dementia, possible, probable), clinical experience teaches us that functional trajectories diverge markedly across individuals with similar test scores but different underlying pathologies.

For instance, individuals with predominant vascular contributions to cognitive impairment (VCI) often exhibit executive dysfunction and apathy early in the disease course, potentially manifesting as rigid, stereotyped behavioral patterns with low environmental reactivity.² In contrast, early Alzheimer's disease may present with preserved routine variability but degraded memory recall and temporal disorientation.³ If activity variability truly reflects an individual's capacity to adapt behaviorally in real time, then grouping all types of cognitive impairment together without distinguishing their causes may obscure important differences in underlying mechanisms – ultimately reducing the metric's usefulness in clinical decision-making.

To better integrate activity variability into neurological assessment, we suggest future work link this measure with domain-specific cognitive performance and neuroimaging markers. For example, variability patterns could be examined in relation to frontal-subcortical network integrity via diffusion tensor imaging or task-based functional MRI. Alternatively, clustering patients based on variability signatures and comparing cognitive domain profiles (e.g., attention, planning, visuospatial processing) might help isolate phenotypes with differential progression risks or responsiveness to intervention.

Moreover, the strong correlation observed between activity variability and gait speed raises the question of whether variability reflects cognitive control, motor capacity, or both. Given that physical performance and neural degeneration often co-occur in aging, disentangling the cognitive versus biomechanical determinants of variability may enhance its specificity as a biomarker.

In closing, we commend the authors for advancing the measurement of behavioral complexity in aging populations. To fully realize its potential, we propose embedding activity variability within a multidimensional framework that accounts for neurobehavioral phenotypes, motor-functional capacity, and underlying neural substrates. Such integration could facilitate more precise risk stratification and ultimately inform individualized care strategies in cognitive neurology.

Writing – original draft: Hui Guo, Ziyu Yang. Writing – review and editing: Xiongfei Zhao.

The authors declare no conflicts of interest. Author disclosures are available in the Supporting Information.