Abstract
INTRODUCTION: Early detection of objective cognitive impairment is essential to delay or prevent dementia; however, traditional in-person screening often faces practical barriers, including limited accessibility and substantial personnel demands. Web-based cognitive tools are promising for scalable screening. This study aimed to identify cognitive subgroups among community-dwelling older adults using latent class analysis (LCA) based on data collected through a freely accessible web-based cognitive screening platform that enables convenient participation anytime and anywhere using older adults' own mobile devices. METHODS: Between September and December 2024, adults aged ≥65 from Sapporo and Ebetsu, Japan, were recruited via newspaper insert flyers (92,290 households) and community posters. QR codes linked to the study website were optimized for various devices. After obtaining electronic consent, participants completed web-based demographic surveys and cognitive assessments of memory, attention, and processing speed. Subjective health and memory complaints were recorded. LCA identified cognitive subgroups based on performance, complaints, and sociodemographic factors. RESULTS: Among the 528 participants (mean age = 71.2; 57% female), most reported good health (86%) and daily conversations (92%). Cognitive function was generally preserved. LCA revealed four clusters: socially isolated females with high subjective memory complaints (SMCs); cohabiting males with high SMCs; cohabiting females with high health perception and preserved cognition; and older adults with cognitive decline. DISCUSSION: The combination of mass outreach and web-based screening is feasible and effective in identifying diverse cognitive profiles. These findings highlight the mismatch between subjective and objective cognition and the relationship between social context, supporting scalable, tailored approaches and cognitive health.