A Multicenter Smartphone-Based Multimodal Database with Temporally Coupled Cognitive and Behavioral Data for Cognitive Health Research

Background: Cognitive impairment and dementia impose increasing health and societal burdens, highlighting the need for scalable and objective approaches to characterize cognitive health across early identification and longitudinal monitoring. Traditional neuropsychological assessments have limited scalability and capture coarse performance outcomes, while digital technologies enable scalable multimodal assessment integrating cognitive, behavioral, and biological data. Accordingly, the Chinese mobile Brief Cognitive Test—Multimodal Integrated Neurocognitive Database (CmBCT-MIND) was established as a multicenter, smartphone-based multimodal database that integrates temporally coupled cognitive and behavioral data with complementary validated scales and biological measures for large-scale cognitive health research. Methods: This study represents the baseline phase of an ongoing longitudinal project initiated in July 2024. Participants without a history of diagnosed cognitive impairment were enrolled from three centers in Beijing. Each participant completed a clinical case report form, validated scales, and the Chinese mobile Brief Cognitive Test (CmBCT). Complementary molecular measures were obtained through fasting venous blood collection for Alzheimer’s disease (AD) biomarker testing (p-tau217, Aβ40, Aβ42) and 5-hydroxymethylcytosine (5hmC) sequencing. A subsample of participants also completed CmBCT with a professional eye-tracking device to capture gaze dynamics during cognitive tasks. Results: The current release of the CmBCT-MIND includes data from 682 participants, covering seven major domains: demographic information, validated scales, cognitive performance from the CmBCT, task-related parameters recorded during CmBCT, time-synchronized behavioral data including touch-interaction features and front-facing videos, molecular data from 481 participants, and eye-tracking recordings from 154 participants. Representative CmBCT-derived task metrics demonstrated expected group-wise differences between Normal and MCI participants, consistent covariate-adjusted associations with MoCA scores, stable effect directions across split samples, and moderate discriminative performance for MCI (AUCs 0.64–0.76). Conclusion: The CmBCT-MIND provides a comprehensive, standardized, and scalable resource for investigating cognitive health and digital biomarker discovery in aging populations. It links smartphone-based cognitive performance with temporally coupled behavioral features, complemented by demographic, neuropsychological, and biological data. The project serves as an evolving open platform that invites collaboration and secondary analyses to advance digital cognitive health research.