·¬ÇÑÉçÇø

Topic

Modeling the Latent Factor Structure of Gait in Healthy Older Adults Using Exploratory Structural Equation Modeling

Department of Psychology

Date & location

• Thursday, August 1, 2024
• 10:00 A.M.
• Virtual Defence

Examining Committee

Supervisory Committee

• Dr. Stuart MacDonald, Department of Psychology, ·¬ÇÑÉçÇø (Co-Supervisor)
• Dr. Jim Tanaka, Department of Psychology, UVic (Co-Supervisor)

External Examiner

• Dr. Sandra Hundza, School of Exercise Science, Physical and Health Education, UVic

Chair of Oral Examination

• Dr. Adam Murray, Department of Computer Science, UVic

Abstract

Objective: This thesis examined the relationship between gait dynamics and cognitive functioning in older adults, focusing on specific aspects such as pace/rhythm, variability, and asymmetry using Exploratory Structural Equation Modeling (SEM). The study aimed to replicate the latent factor model of gait by Arcolin et al. (2022), refine it using Exploratory SEM, and explore its criterion validity as a predictor of individual differences in executive functioning. The innovation in Arcolin et al.’s (2022) model lies in its multivariate approach, integrating multiple gait parameters to better capture the complexity of gait performance. The model indexes three latent factors (pace/rhythm, variability, and asymmetry) using eight gait indicators.

Methods: This study used archival baseline data from the Healthy Bodies, Healthy Minds (HBHM) program, involving 115 low-active community-dwelling older adults aged 65-88 years (M =72.82, SD = 5.25). Gait patterns were measured using the GAITRite system, which captures both spatial and temporal aspects of gait. Eight key gait indicators were analyzed: gait speed, step time, double support time, step length CV, swing time CV, step velocity CV, step time asymmetry, and step length asymmetry. SEM was employed to address three main questions: (1) replication of the Arcolin et al. (2022) latent factor model of gait using Confirmatory Factor Analysis (CFA), (2) identification of the optimal latent factor structure of gait in older adults using Exploratory SEM, and (3) development of a structural model to predict individual differences in executive functioning as measured by the Groton Maze Learning Task.

Results: The replication of the Arcolin et al. (2022) model using CFA showed a moderate fit, indicating some limitations in the original model. The Exploratory SEM approach led to a refined model with a significantly improved fit, suggesting a better representation of the underlying gait constructs. The new model, which included step length asymmetry as a spatial metric, demonstrated good mode fit (CFI = 0.959, TLI = 0.933, RMSEA = 0.084), capturing both temporal and spatial aspects of gait asymmetry. The structural model revealed that the gait variability latent factor approached significance in predicting executive functioning (β = 0.26, p =.09), while pace/rhythm and asymmetry did not significantly predict cognitive performance (p > .05).

Conclusions: Evidence for the interdependence between gait performance and cognitive function in older adults was found, particularly highlighting the significance of pace/rhythm. The use of Exploratory SEM provided a detailed understanding of the complex relationships between various gait indicators, improving the accuracy of latent factor models of gait models. These findings have important implications for early detection of cognitive decline and fall risk in older adults. Robust latent factor models of gait can enhance early identification of gait abnormalities, enabling timely interventions to prevent falls and improve cognitive health outcomes. This approach could lead to better clinical assessments and personalized interventions, ultimately helping older adults maintain their independence and quality of life.