Objective: To determine if digital mobility outcomes (DMOs) differentiate between those with and without cognitive impairment in Parkinson’s disease (PD).

Background: Cognitive impairment is common and impacts the quality of life in people with PD (PwPD). In-clinic assessed gait outcomes have been shown to discriminate between people with and without cognitive impairment. However, this still requires in-clinic visits whereas real-world DMOs provide continuous monitoring remotely and may have more practical utility.

Method: As part of the international Mobilise-D study, 601 PwPD were asked to wear a single wearable sensor on the lower back for a 7-day period, enabling 24 technically validated DMOs. Cognition was assessed using the Montreal Cognitive Assessment score (MoCA). PwPD were split into 2 groups based on MoCA score (cognitively normal: MoCA ≥26/30 vs. cognitively impaired: MoCA <26/30). Area under the curve (AUC) from (10-fold cross-validation) logistic regression was used to investigate the discriminative ability of each DMO combined with demographic covariates age, gender, height, weight, education years and disease duration. Next, an elastic net (inherently using a 10-fold cross-validation, α=0.5) penalized logistic regression model was used for optimal variable selection and to assess if combined DMOs improve cognitive status classification.

Results: Valid DMO and MoCA data were available for 320 cognitively normal and 209 cognitively impaired PwPD. The best discriminating DMO was the percentile 90 walking speed in walking bouts longer than 30 seconds (AUC = 0.668). However, combined with the demographic covariates, AUC increased to 0.761. Furthermore, the demographics alone achieved an AUC of 0.750. The elastic net model selected age, educational years, percentile 90 walking speed in >30 seconds walking bouts and mean walking speed in walking bouts between 10 and 30 seconds (AUC = 0.735).

Conclusion: Real-world gathered percentile 90 walking speed during >30 seconds walking bouts can differentiate cognitively impaired PwPD from non-impaired, but demographic information must be included too. These findings underscore the potential of real-world DMOs as part of a pre-screening battery for further in-depth cognitive testing.

« Back to 2026 International Congress

MDS Abstracts - https://www.mdsabstracts.org/abstract/can-digital-mobility-outcomes-differentiate-cognitive-impairment-in-parkinsons-disease/