Web-Based Application for Cognitive and Functional Assessments in Dementia Screening: Mixed Methods, User-Centered Development Approach

Background: Digital health technologies offer new opportunities for cognitive screening and monitoring among older adults. In Thailand, where dementia prevalence is rising, accessible web-based cognitive tools remain limited despite their potential to facilitate early detection and community-based assessment. Understanding usability and validity is critical to ensure successful implementation in real-world contexts. Objective: This study aimed to develop and validate a web-based application, Healthy Brain Test, for cognitive and functional assessments in dementia screening among older Thai adults. Specific objectives were to (1) design user-centered cognitive modules covering key cognitive domains and (2) evaluate correlations between the web-based assessments and conventional clinical tools to determine diagnostic cutoffs for cognitive impairment. Methods: We designed Healthy Brain Test as a self-administered web application suitable for older users and their caregivers. The platform includes digital versions of the Thai Mental State Examination (e-TMSE), a clock drawing test, and a category verbal fluency test, along with electronic versions of the short form of the Informant Questionnaire on Cognitive Decline in the Elderly (IQCODE-16) and cognitive instrumental activities of daily living (IADLs). Participants completed both web-based and paper-based assessments. Correlations between modalities were analyzed, and receiver operating characteristic (ROC) curves were generated to determine sensitivity and specificity. Data were analyzed using SPSS for Windows, version 30.0 (IBM Corp) and MedCalc Statistical Software (MedCalc Software Ltd). Results: A total of 198 older adults participated (women: 137/198, 69.2%; median age 69.4 years), with 57.1% (113/198) having more than 6 years of education. Of the 198 participants, 44 were diagnosed with major neurocognitive disorder, 58 were diagnosed with mild neurocognitive disorder, and 96 were cognitively normal. The e-TMSE showed strong agreement with the traditional TMSE (r=0.837; P