Effects of augmented reality cueing strategies on freezing of gait: The ELIMINATE FoG trial

Abstract

Background: Freezing of gait (FoG) is a treatment-resistant symptom of Parkinson disease (PD). Augmented reality (AR) cues have been investigated as a therapy for FoG, with inconclusive results from a limited array of AR constructs.

Objectives: Compare four modalities of a novel AR cue to physical and no-cue controls.

Methods: Presence of FoG in PD was required; exclusion criteria included dementia, severe vision loss, and significant gait-disrupting comorbidities. Participants completed six walking tasks, featuring different cueing conditions in a crossover fashion, in a holographic hallway displayed by an AR headset. A conventional physical cue was presented first, followed by other conditions in randomized order (hand-controlled AR cue, observer-controlled AR cue, eye-controlled AR cue, constant AR cue, no-cue control). Primary outcomes were FoG duration and incidence, manually annotated. Secondary outcomes included survey questions and gait parameters derived from IMUs.

Results: Thirty-six participants completed testing. The observer-controlled AR cue produced lower FoG duration than the no-cue, physical, and hand-controlled AR cue conditions (N = 36, p ≤ 0.006, Wilcoxon effect size (WES) ≥ 0.46). The constant cue reduced FoG incidence compared to all other conditions (N = 36, p ≤ 0.016, WES ≥ 0.40). Participants' preferred AR cues decreased FoG duration (N = 28, p ≤ 0.004, WES ≥ 0.48) and incidence (N = 28, p ≤ 0.022, WES ≥ 0.38) compared to controls. Differences in kinematic outcomes were negligible. Survey results indicated receptiveness toward AR cueing, with diversity in preferred cue activation modalities. No significant adverse events occurred.

Conclusions: AR cueing decreased FoG incidence and duration compared to controls. Efficacy of discrete cueing modalities likely depends on user intrinsic factors, such as preference.

Keywords: Augmented reality; Cue; FoG; Gait freezing; Parkinson’s disease.

Fig. 1

Aerial (A) and lateral (B) views of the novel AR cue, and 3D models of course Level 1, 2 (C) and 3 (D) from aerial view. Green traces represent the paths of two participants throughout all six conditions. Thick black traces indicate FoG episodes. Blue circles represent the pivot circle. Columns and tall walls were 2.5 m high. Short walls were 0.5 m high. The interior width of the doorways was 0.55 m, and the distance between columns was 0.5 m. The hallway itself was 7.5 m long and 2.2 m wide.

Fig. 2

Percent time frozen (top) and freeze rates (bottom), per condition, ordered by descending medians. Significant differences in the conditions are denoted with brackets, with significance based on Wilcoxon signed-rank tests.