Mapping Patient-Reported Functional Mobility and Digital Mobility Outcomes in Parkinson’s Disease#

Authors#

Alan Castro Mejia, Stefano Sapienza, Patricia Martins Conde, Jean-Paul Steinmetz, Lukas Pavelka, Rejko Krüger, Jochen Klucken

Abstract#

Background: Mobility is a cornerstone of neurological assessment, influencing overall health and quality of life. While subjective self-reported mobility complements clinical evaluations, current visual assessments often fail to account for the patient’s perception or the complexities of daily activities. Objective measures such as digital gait outcomes may bridge this gap.

Aim: To determine which gait parameters best align (correlate) with the patient’s perception of functional mobility (PRM).

Methods: Cross-sectional data from 94 participants with Parkinson’s Disease extracted from the Luxembourg Parkinson’s Study. Gait outcomes were derived from three tasks of increasing complexity based on Time-Up and Go (TUG). Multivariate regression models were defined using Least Absolute Shrinkage and Selection Operator to determine the best predictors of PRM. Calibration, beta regression, and quantile regression were used to enhance clinical interpretability.

Results: Regression models integrating clinical and gait data demonstrated improved PRM prediction accuracy, with the TUG (R2: 0.35) providing the most comprehensive insights. Beta regression indicated that higher PRM was independently associated with higher gait speed, lower medication use, lower depressive symptoms, and absence of clinician-reported mobility impairment. Translating clinically to a +0.10 m/s increase in gait speed corresponding to +1.39 patient-reported functional mobility points.

Conclusions: Instrumented TUG assessments capture gait parameters linked to how patients perceive their functional mobility. These findings support the targeted use of DMOs—particularly gait speed and swing-phase metrics—to complement clinical evaluation, especially in patients reporting poorer mobility. Longitudinal studies and external validation cohorts are needed to determine the stability and generalizability of the models.