Performance of a Personalized Smart Cueing Device to Detect Freezing of Gait in Parkinson's Disease.
Performance of a Personalized Smart Cueing Device to Detect Freezing of Gait in Parkinson's Disease.
Where did the research take place?
The study site has not been established. Author addresses may differ from where the research occurred.
Amsterdam, NL · Author affiliation
Department of Rehabilitation Medicine, Amsterdam UMC, Amsterdam, the Netherlands.Location evidence
The Hague, NL · Author affiliation
Cue2Walk International B.V, The Hague, the Netherlands.Location evidence
Nijmegen, NL · Author affiliation
Donders Institute for Brain, Cognition and Behaviour; Department of Rehabilitation; Centre of Expertise for Parkinson & Movement Disorders, Radboud University Medical Centre, Nijmegen, the Netherlands.Location evidence
A plain-language reading has not been prepared for this paper yet.
Original abstract
External cueing is a proven strategy to overcome freezing of gait (FoG) in Parkinson's disease (PD). However, manual activation of cueing is often too late and continuous cueing may be perceived as burdensome. Therefore, the Cue2walk, a medical device for FoG detection and provision of external cues, was developed. This study aimed to determine (1) the performance to detect FoG of the Cue2walk device and (2) the effect of personal optimization on its performance to detect FoG. Twenty-four people with PD and daily FoG completed a gait circuit including FoG-triggering tasks in their own home environment, while wearing a Cue2walk device to collect acceleration data. Two experienced independent raters rated the occurrence of FoG using video annotation. The data set of each participant was split into a training (60%) and test set (40%), obtaining an even distribution of FoG episodes. The training set was used to personalize the settings of the algorithm. The test set was subsequently used to assess the performance to detect FoG of the standard and personalized setting by calculating sensitivity, specificity, and latency (time between onset and detection of FoG). Sensitivity was 28.2% ± 31.6%, specificity was 96.0% ± 4.1%, and latency was 1.9 s ± 2.4 s on the test set with the standard settings. With personalized settings, sensitivity was 89.1% ± 14.8%, specificity was 93.3% ± 5.5%, and latency was 1.6 s ± 2.2 s. Sensitivity was significantly higher for the personalized settings (p < 0.001), while latency and specificity remained similar. The Cue2walk device showed good performance to detect FoG when using the personalized settings. The clinical effectiveness needs to be determined in future research.