Related Experiment Video
Updated: Apr 19, 2026
Video Movement Analysis Using Smartphones ViMAS: A Pilot Study
Published on: March 14, 2017
Smartphone-based gait analysis with rhythmic auditory cueing discriminates cognitive impairment stages in a
Samwel Sylvester Msigwa1,2, Innocent Rwiza Mwombeki2, Sadiki Mandari2
1The Second Affiliated Hospital and Yuying Children's Hospital of Wenzhou Medical University, Department of Neurology, Institute of Geriatric Neurology, Wenzhou, Zhejiang, China.
Background:
Early cognitive impairment detection in sub-Saharan Africa is limited by few specialists, restricted testing and cultural barriers. Conventional dual-task paradigms may be unsuitable for low-literacy populations, whereas rhythmic auditory cueing (RAC) using familiar music may provide a culturally neutral, engaging alternative. This study aimed to validate smartphone-based gait assessment across cognitive stages in adults attending a memory-clinic.
Methods:
Sixty adults aged 60-85 years attending a memory clinic were stratified into Subjective Cognitive Decline (SCD, n = 21), Mild Cognitive Impairment (MCI, n = 23) and Dementia (n = 16). Smartphone-based spatiotemporal and kinematic gait parameters were collected under uncued, dual-task (fruits/vegetables naming) and RAC conditions. Between-group differences were examined using MANCOVA or rank-based MANOVA, adjusting for age, sex and body mass index, with Bonferroni-corrected post hoc tests. Correlations with executive function were analysed using false discovery rate-corrected Spearman tests. Receiver operating characteristic analyses determined discriminatory performance.
Results:
Gait impairment increased with cognitive decline. Dementia participants exhibited slower velocity, shorter stride length, longer double support, reduced hip range and higher stance asymmetry, particularly under RAC. RAC stance asymmetry distinguished SCD from Dementia with high accuracy (AUC = 0.952, sensitivity 93.8%, specificity 90.5%), outperforming dual-task measures. Dual-task base width differentiated SCD from MCI (AUC = 0.772), while music-cued support times distinguished MCI from Dementia (P ≤ .028, η2 ≈ 0.33). Gait parameters correlated strongly with executive function (r ≈ 0.48, P < .001).
Conclusions:
Smartphone-based gait assessment, particularly under RAC, provides a sensitive, culturally neutral and scalable method for detecting and classifying cognitive impairment. RAC highlights sensorimotor deficits characteristic of dementia, complementing standard cognitive screening tools in resource-limited settings.
