JMIR Publications

ABSTRACT

Background:

Cognitive impairment affects millions of older adults worldwide, with Alzheimer's disease and mild cognitive impairment posing significant public health challenges. Early identification is critical for timely intervention, yet traditional cognitive assessment tools such as the MMSE and MoCA have inherent limitations, including educational bias and ceiling effects, while biomarker-based approaches remain costly and invasive. In recent years, the rapid advancement of multimodal data fusion technologies—integrating physiological signals, behavioral data, speech, and environmental information from wearable devices, sensors, and smartphones—has opened new possibilities for continuous, non-invasive cognitive assessment in community and home settings. However, despite growing interest in this field, a systematic synthesis of how multimodal data are being applied specifically in care for older adults with cognitive impairment is currently lacking. This scoping review aims to map current data types, collection methods, and application contexts, and to identify gaps in the evidence base to inform future research and practice.

Objective:

To conduct a scoping review on the application of multimodal data in care for older adults with cognitive impairment, identifying current data types, collection methods, and application contexts, and to provide evidence-informed insights for future research and practice.

Methods:

Following the Joanna Briggs Institute methodology for scoping reviews, we systematically searched five English databases: PubMed, Web of Science, Embase, Scopus, and CINAHL, from inception to July 2026. Additionally, we searched ClinicalTrials.gov and the WHO International Clinical Trials Registry Platform (ICTRP) to identify ongoing or unpublished clinical trials, and manually screened the reference lists of included studies. Original studies were included if they involved the collection, fusion, or analysis of two or more data modalities; focused on individuals aged 50 years and older with cognitive impairment; and addressed cognitive assessment, screening, monitoring, or care management contexts. Thematic analysis was conducted to synthesize the extracted data.

Results:

A total of 16 studies published between 2021 and 2026 were included. Multimodal data types were summarized into seven categories (a–g): physiological and sleep data (Category a), behavioral and motor data (Category b), multimedia data (Category c), linguistic and textual data (Category d), environmental data (Category e), clinical data (Category f), and patient-reported outcomes (Category g). Data collection methods encompassed various sensors (wearable devices, environmental sensors, vision-based sensors, in-vehicle sensors, etc.), extraction from hospital information systems, and scale-based assessments. Application contexts primarily included cognitive assessment and early screening, remote monitoring and home-based care, and behavioral symptom management in inpatient and long-term care settings.

Conclusions:

Multimodal data can effectively enhance the accuracy of cognitive assessments and the timeliness of care decisions in older adults with cognitive impairment. However, research in this field remains in its early stages, with the current evidence predominantly derived from cross-sectional studies, limited sample sizes, and insufficient model interpretability. Future research should prioritize longitudinal designs, multicenter large-sample validation, and improved model interpretability to promote the standardized application and clinical translation of multimodal data in cognitive impairment care.