Mapping 15 Years of Wearable Gait Research for Spotting Early Dementia
A bibliometric review of 126 studies shows wearable gait sensors are a growing, multidisciplinary route to early dementia detection, but validation gaps remain.
Summary
Walking changes can appear early in cognitive decline, and wearable sensors can measure them cheaply and non-invasively. This bibliometric review searched Scopus for 2010–2025 studies on wearable sensors and gait analysis for early dementia detection and analyzed 126 papers with VOSviewer. The field spans neurology, geriatrics, engineering, and computer science, led by the United States and its collaborators in Europe. Four themes emerged: cognitive and clinical aspects of dementia, physical activity and mobility in older adults, technology and analytics for gait and frailty, and aging with emerging technologies such as machine learning. The authors note gaps in long-term validation, standardized methods, and integration into clinical workflows. The paper maps the research landscape rather than testing diagnostic accuracy.
Detailed Summary
Dementia affects over 55 million people worldwide, a number projected to reach 139 million by 2050, and up to 60% of cases may go undiagnosed in primary care. Neuroimaging and cerebrospinal fluid biomarkers are effective but costly, invasive, and often inaccessible. Gait depends on attention, executive function, and memory, and it can change subtly in early cognitive decline. That makes walking a candidate digital biomarker, measurable with wearable inertial sensors or accelerometers.
The authors ran a bibliometric review rather than a clinical study. They searched Scopus for English-language original articles on human participants published 2010–2025. The query combined dementia/cognitive impairment terms, detection or monitoring terms, gait or mobility terms, and sensor terms. Of 184 records retrieved, 126 met the criteria. They used VOSviewer to run performance analysis (authors, journals) and science mapping (co-authorship, bibliographic coupling, co-citation, and keyword co-occurrence).
The most-cited authors included Jeffrey Kaye (563 citations from 6 papers), Nora Mattek, Tamara Hayes, and Bijan Najafi. Leading journals were Sensors, Journal of Alzheimer's Disease, and Gait & Posture, with notable output also in IEEE Journal of Biomedical and Health Informatics, Gerontology, and Journal of Medical Internet Research. The United States sat at the center of the collaboration network, closely linked with the UK, Italy, Germany, and the Netherlands. Co-citation analysis showed a gradient from dementia and geriatrics journals through neurology and aging science to general clinical medicine. Keyword analysis identified four themes: cognitive and clinical aspects of dementia; physical activity and mobility in older adults; technological and analytical approaches to gait and frailty (deep learning, sensors); and aging, cognitive decline, and emerging technologies (machine learning, MCI).
The discussion in the available text highlights dual-task cognitive-motor assessments as promising. One example is a Performance Index combining gait and cognitive accuracy during an instrumented Timed Up and Go test, which strongly predicted MMSE scores. Others include a Dual Task Ball Balancing Test linked to MCI and Alzheimer's-related dementia, and greater minimum toe clearance variability in MCI. Poorer balance and gait also predicted falls and wheelchair dependence.
The authors conclude that sensor-derived gait metrics could strengthen early diagnostic pathways, but they identify gaps in longitudinal validation, methodological standardization, and clinical integration. Because this is a bibliometric map, it shows where research is concentrated, not how accurate gait sensors are for diagnosis. The provided full text is also truncated in the discussion, and the single-database search may have missed relevant studies.
Key Findings
- 126 peer-reviewed studies (2010–2025) on wearable gait analysis for early dementia detection were identified from 184 Scopus records.
- Jeffrey Kaye was the most-cited author (563 citations from 6 papers); Sensors, J Alzheimer's Disease, and Gait & Posture led journal output.
- The US anchors a collaboration network linked with the UK, Italy, Germany, and the Netherlands.
- Four thematic clusters emerged, spanning clinical dementia, mobility in older adults, sensor/deep-learning analytics, and aging with machine learning.
- Dual-task cognitive-motor tests, such as the instrumented TUG Performance Index, showed promise, but longitudinal validation and standardization remain lacking.
Methodology
Bibliometric review of Scopus-indexed English-language original human studies from 2010–2025; 184 records were screened and 126 included. VOSviewer 1.6.20 was used for performance analysis plus co-authorship, bibliographic coupling, co-citation, and keyword co-occurrence mapping.
Study Limitations
The review maps publication patterns and does not pool or assess diagnostic accuracy or study quality. It relied on a single database (Scopus), English-only articles, and a restrictive 'Humans' keyword filter, so relevant work may be missed. The provided text is truncated in the discussion, and the cluster and threshold descriptions are internally inconsistent in places.
Enjoyed this summary?
Get the latest longevity research delivered to your inbox every week.
Enter your email to subscribe:
