Abstract
<title>Abstract</title> <p>With global population aging, cognitive impairment has become a major public health challenge. Current assessment approaches provide important information on cognitive function and disease-related pathology, but offer limited characterization of task-evoked physiological and behavioral responses associated with cognitive impairment. Here, we investigated task-evoked physiological and behavioral responses using a wearable sensing system that synchronously records electroencephalography, electrocardiography, speech, and video signals during structured cognitive tasks. A total of 1,385 community-dwelling adults aged 60 years or older were recruited from five administrative districts in Nanjing, China, and divided into an internal development cohort (n=831) and an external validation cohort (n=554). The system distinguished normal cognition from cognitive impairment, with areas under the receiver operating characteristic curve of 0.93 and 0.90 in the internal and external cohorts, respectively. These findings demonstrate the utility of wearable multimodal signals for accurate assessment of cognitive impairment. Importantly, different cognitive tasks elicit distinct physiological-behavioral features that contribute to cognitive impairment assessment and reflect distinct, complementary cognitive functions. These findings suggest that task-evoked wearable multimodal signals can provide process-level functional phenotyping information for cognitive impairment assessment, complementing conventional cognitive assessment.</p>