Associations of eye-tracking measures with global cognitive performance in community-dwelling older adults: an exploratory cross-sectional study

Eye tracking has been investigated as an instrumentally measured approach to characterizing cognitive performance in older adults. As an initial step toward the longer-term investigation of low-burden repeated monitoring, we examined whether selected eye-tracking measures were associated with global cognitive performance in community-dwelling older adults. In this exploratory cross-sectional study, 38 community-dwelling older adults completed neuropsychological tests and a brief eye-tracking battery. The main analyses examined associations of area-of-interest viewing proportion (AOI%) and antisaccade latency with continuous MoCA and MMSE-J scores using Spearman correlations with Benjamini–Hochberg false discovery rate control. Secondary exploratory analyses examined separation between groups defined by relatively lower and higher cognitive screening scores; these groups were not considered clinical diagnostic categories. AOI% was positively associated with global cognitive scores in the primary analyses (MoCA ρ = 0.489, q = 0.004; MMSE-J ρ = 0.520, q = 0.003). Antisaccade latency was negatively associated with MMSE-J (ρ = −0.414, q = 0.013), while its association with MoCA was weaker. In exploratory discrimination analyses, AOI% showed AUCs of 0.780 for the lower-MoCA group (< 26) and 0.763 for the lower-MMSE-J group (≤ 27), while antisaccade latency showed AUCs of 0.672 and 0.785, respectively. In exploratory incremental models, optimism-corrected AUCs across Models 1–3 were 0.633, 0.769, and 0.831 for the lower-MMSE-J group and 0.664, 0.764, and 0.752 for the lower-MoCA group. AOI-based gaze allocation and antisaccade latency showed preliminary associations with continuous global cognitive performance in this community-based sample. The findings support further investigation of brief eye-tracking measures but do not establish early detection, sensitivity to longitudinal cognitive change, diagnostic accuracy, home-based feasibility, or incremental clinical utility beyond MoCA or MMSE-J. Larger longitudinal studies using clinically characterized outcomes and externally validated measurement systems are required.