Validation of the mobgap analytical pipeline for estimation of digital mobility outcomes from a single wearable device across all-cause dementia and discrete dementia disease subtypes

Abstract Background Wearable-derived digital mobility outcomes have potential clinical applications in dementia, yet technical validation of analytical pipelines in this population remains limited. This study evaluated the validity of the mobgap pipeline for estimating walking speed, initial contact detection, stride duration, cadence, and stride length across all-cause dementia and discrete subtypes. Methods Eighty-six participants were recruited from two studies (GaitDem: Alzheimer’s Dementia (AD) = 31, Dementia with Lewy-bodies (DLB) = 25, Parkinson’s Dementia (PDD) = 8, Vascular Dementia (VaD) = 6; UCL DMS Atypical: posterior cortical atrophy (PCA) = 16), completing laboratory walking trials validated against an instrumented walkway (GaitDem) or infrared motion capture (DMS Atypical). DMOs were estimated using the mobgap pipeline, optimised for impaired walking. Results Walking speed was estimated with good agreement across all-cause dementia (Intraclass correlation coefficient (ICC) = 0.83, absolute error = 0.12 m/s, relative error = 11.84%), comparable to previous laboratory validations in gait-impaired populations. Cadence showed the strongest performance (ICC = 0.98, absolute error = 1.34 s/m, relative error = 1.24%), while stride length was the most challenging (ICC = 0.77, relative error = 11.85%). Initial contact detection achieved high sensitivity (0.99) but moderate positive predictive value (PPV) (0.85). By subtype, ICC for walking speed ranged from 0.76 (AD) to 0.89 (VaD), with PDD presenting the greatest challenge across most digital mobility outcomes, likely reflecting motor impairment severity rather than dementia subtype. Conclusion These findings support the technical validity of mobgap for estimating digital mobility outcomes across common and atypical dementia subtypes, and highlight the need for real-world validation to establish its clinical utility.

Authors

Institutions

Publication Details

Journal
Journal of NeuroEngineering and Rehabilitation
Published
2026-10-31
DOI
https://doi.org/10.1186/s12984-026-02198-x
Primary Topic
Balance, Gait, and Falls Prevention
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
article

Validation of the mobgap analytical pipeline for estimation of digital mobility outcomes from a single wearable device across all-cause dementia and discrete dementia disease subtypes

Keir X. X. Yong, Alan J. Thomas, Diego N. Kaski, Cameron Kirk et al.
Journal of NeuroEngineering and Rehabilitation
Balance, Gait, and Falls Prevention
article

Validation of the mobgap analytical pipeline for estimation of digital mobility outcomes from a single wearable device across all-cause dementia and discrete dementia disease subtypes

Keir X. X. Yong, Alan J. Thomas, Diego N. Kaski, Cameron Kirk, Silvia Del Din, Lynn Rochester, Matthew James Bancroft, Lisa Alcock, Brook Galna, Ríona Mc Ardle
article en

Abstract

Abstract Background Wearable-derived digital mobility outcomes have potential clinical applications in dementia, yet technical validation of analytical pipelines in this population remains limited. This study evaluated the validity of the mobgap pipeline for estimating walking speed, initial contact detection, stride duration, cadence, and stride length across all-cause dementia and discrete subtypes. Methods Eighty-six participants were recruited from two studies (GaitDem: Alzheimer’s Dementia (AD) = 31, Dementia with Lewy-bodies (DLB) = 25, Parkinson’s Dementia (PDD) = 8, Vascular Dementia (VaD) = 6; UCL DMS Atypical: posterior cortical atrophy (PCA) = 16), completing laboratory walking trials validated against an instrumented walkway (GaitDem) or infrared motion capture (DMS Atypical). DMOs were estimated using the mobgap pipeline, optimised for impaired walking. Results Walking speed was estimated with good agreement across all-cause dementia (Intraclass correlation coefficient (ICC) = 0.83, absolute error = 0.12 m/s, relative error = 11.84%), comparable to previous laboratory validations in gait-impaired populations. Cadence showed the strongest performance (ICC = 0.98, absolute error = 1.34 s/m, relative error = 1.24%), while stride length was the most challenging (ICC = 0.77, relative error = 11.85%). Initial contact detection achieved high sensitivity (0.99) but moderate positive predictive value (PPV) (0.85). By subtype, ICC for walking speed ranged from 0.76 (AD) to 0.89 (VaD), with PDD presenting the greatest challenge across most digital mobility outcomes, likely reflecting motor impairment severity rather than dementia subtype. Conclusion These findings support the technical validity of mobgap for estimating digital mobility outcomes across common and atypical dementia subtypes, and highlight the need for real-world validation to establish its clinical utility.

Journal of NeuroEngineering and Rehabilitation
University College London Hospitals NHS Foundation Trust (GB), Murdoch University (AU), Newcastle upon Tyne Hospitals NHS Foundation Trust (GB), National Institute for Health and Care Research (GB), NIHR Newcastle Biomedical Research Centre (GB), UCL Queen Square Institute of Neurology (GB), University College London (GB), Newcastle University (GB)
Openalex Percentile: Top 5%
Balance, Gait, and Falls Prevention
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon