Comment on “Eye‐Tracking‐Based Cognitive Assessment Predicts the Risk of Memory Decline: A Community‐Based Cohort Study”

We read with great interest the study by Katsuhisa et al. [1], which evaluates the prognostic potential of an eye-tracking-based cognitive assessment (ETCA) for predicting future memory decline in a community cohort. The investigation addresses an important gap in dementia prevention strategies, as scalable tools capable of detecting subtle cognitive vulnerability before overt impairment remain limited. The authors present an innovative digital approach that can be completed within minutes and that appears capable of detecting early cognitive differences not captured by conventional screening instruments. One aspect that warrants further consideration relates to the statistical robustness and interpretability of the predictive performance reported for ETCA. The receiver operating characteristic analysis demonstrated an area under the curve (AUC) of 0.709 for the composite subscores, indicating moderate discrimination rather than strong predictive accuracy. In clinical risk stratification contexts, models with AUC values around 0.70 often require additional calibration analysis and decision-curve evaluation to determine whether they meaningfully improve patient classification [2]. Without these assessments, it remains uncertain whether ETCA-derived risk estimates would alter screening decisions or clinical management pathways. Integrating calibration metrics, such as calibration plots or Brier scores, and exploring clinically relevant decision thresholds would help clarify whether the observed statistical discrimination translates into actionable prognostic utility. Another interpretive challenge concerns the reliance on a single cognitive outcome measure to define future decline. In the present study, memory deterioration was operationalized solely through changes in the Rivermead Behavioral Memory Test (RBMT). Although RBMT is sensitive to everyday memory performance, cognitive aging often involves multidomain trajectories involving executive function, attention, and processing speed. A predictive instrument intended for dementia risk stratification may therefore benefit from validation against broader composite outcomes incorporating multiple cognitive domains or functional indicators [3]. Limiting outcome classification to one test could restrict the capacity to capture heterogeneous early trajectories of neurodegeneration, thereby influencing estimates of predictive performance and clinical applicability. The physiological interpretation of the ETCA signals also deserves further clarification. Eye-tracking metrics capture gaze allocation and viewing time within task-defined regions of interest, which may reflect a mixture of attentional control, visuospatial processing, and task engagement. However, the neurocognitive pathways linking these gaze behaviors to future episodic memory decline remain incompletely characterized. Without mechanistic contextualization, it is difficult to determine whether reduced gaze toward correct targets reflects early hippocampal dysfunction, broader attentional inefficiency, or noncognitive factors such as fatigue or visual scanning strategies. Incorporating complementary neuroimaging or biomarker correlations could help delineate whether ETCA signals correspond to established neurodegenerative pathways or represent a distinct digital behavioral phenotype [4]. Finally, the practical translation of ETCA into population-level screening programs would benefit from additional evaluation of implementation factors. Although the authors emphasize the rapid and low-burden nature of the assessment, large-scale deployment requires standardization of hardware calibration, environmental lighting conditions, and participant positioning to ensure reliable gaze tracking across diverse settings. Variability in these operational parameters could influence signal quality and measurement stability, particularly in older populations with ocular comorbidities. Prospective studies examining reproducibility across clinical environments and community screening platforms would therefore strengthen confidence in the tool's scalability [5]. Despite these considerations, the study provides an important contribution to the emerging field of digital cognitive biomarkers. By demonstrating that eye-tracking-derived metrics may identify individuals at elevated risk of subtle cognitive deterioration, the authors offer a promising direction for scalable early detection strategies. Continued methodological refinement and integration with multimodal biomarkers may further clarify the role of ETCA within preventive frameworks aimed at delaying or mitigating dementia onset. Sangameshwar Manikya: validation, writing – original draft, writing – review and editing. Varshini Vadhithala: supervision, project administration, writing – original draft, writing – review and editing. Rajnish Kumar: writing – original draft, writing – review and editing. Pankaj Nainwal: conceptualization, methodology, writing – original draft, writing – review and editing. All authors reviewed and approved the manuscript. During the preparation of this work the authors used Grammarly and ChatGPT in order to refine language, grammar, and style. After using this tool/service, the author(s) reviewed and edited the content as needed and took full responsibility for the content of the publication. The authors have nothing to report. The authors have nothing to report. The authors have nothing to report. The authors declare no conflicts of interest. Data sharing not applicable to this article as no datasets were generated or analyzed during the current study.