Independent domains of daily mobility in patients with neurological gait disorders

Abstract

The aim of this study was to establish a comprehensive and yet parsimonious model of daily mobility activity in patients with neurological gait disorders. Patients (N = 240) with early-stage neurological (peripheral vestibular, cerebellar, hypokinetic, vascular or functional) gait disorders and healthy controls (N = 35) were clinically assessed with standardized scores related to functional mobility, balance confidence, quality of life, cognitive function, and fall history. Subsequently, daily mobility was recorded for 14 days by means of a body-worn inertial sensor (ActivPAL^®). Fourteen mobility measures derived from ActivPAL recordings were submitted to principle component analysis (PCA). Group differences within each factor obtained from PCA were analyzed and hierarchical regression analysis was performed to identify predictive characteristics from clinical assessment for each factor. PCA yielded five significant orthogonal factors (i.e., mobility domains) accounting for 92.3% of the total variance from inertial-sensor-recordings: ambulatory volume (38.7%), ambulatory pattern (22.3%), postural transitions (13.3%), sedentary volume (10.8%), and sedentary pattern (7.2%). Patients' mobility performance only exhibited reduced scores in the ambulatory volume domain but near-to-normal scores in all remaining domains. Demographic characteristics, clinical scores, and fall history were differentially associated with each domain explaining 19.2-10.2% of their total variance. This study supports a low-dimensional five-domain model for daily mobility behavior in patients with neurological gait disorders that may facilitate monitoring the course of disease or therapeutic intervention effects in ecologically valid and clinically relevant contexts. Further studies are required to explore the determinants that may explain performance differences of patients within each of these domains and to examine the consequences of altered mobility behavior with respect to patients' risk of falling and quality of life.

Keywords: Body-worn sensor; Daily mobility; Factor analysis; Gait disorder; Wearable.

Fig. 1

Relative importance of mobility domains and group differences within each domain. a Principle component analysis of the dataset with 14 mobility measures from 275 recordings yielded in total 14 components (black dots) with the five first factors (colored dots) explaining 92.3% of total variance (cumulative explained variance is indicated by gray dots). b Radar plot with median z values (colored dots) and interquartile ranges (gray shaded area) of patients for all five mobility domains. Patient data is normalized with respect to healthy control performance (dotted black line). Daily mobility activity in patients falls within the normal range for all domains except ambulatory volume