Abstract Details

Goal-directed movements (GDMs) play a pivotal role in upper limb motor control, representing planned motor commands directing hand trajectories toward specific targets. Objective monitoring of GDMs during everyday life data offers advantages over traditional methods for assessing upper limb function, providing insights into the quality and complexity of movements. We developed a wearable-based solution that automatically identifies GDMs and extracts GDM characteristics that capture sensitive biomarkers of upper limb health in neurological disorders. 

To assess upper limb function in neurological disorders through continuous monitoring of hand goal-directed movements during activities of daily living using a wrist-worn wearable sensor.

Participants diagnosed with amyotrophic lateral sclerosis (ALS), Huntington's disease (HD), and Friedreich's ataxia (FA) wore a PAMSys ULM™ (BioSensics LLC, Newton, MA USA) wrist sensor for one week after a clinical visit. The deep-learning model detected GDMs, extracting pertinent features characterizing movement quality. Disease-specific clinical assessments (e.g., mFARS, UHDRS, ALSFRS-R) were conducted, and Spearman correlations were performed to examine associations between clinical scores and GDM features.

The study comprised 40 FA participants (average age: 26.7 ± 1.5 years, including 19 females), 25 HD and prodromal HD participants (average age: 41.7 ± 11.2 years, 15 females), and 11 ALS patients (8 males, 3 females; age: 64.4 ± 9.8 years). Strong to moderate correlations were observed between disease-related clinical scores and GDM features, including average daily GDM counts (p=0.008, ρ = 0.576) and entropy of movement elements (p=0.047, ρ = 0.423). In addition, machine learning-based models were developed to predict upper limb health in each of the three disease areas.

Our study presents a wearable-based remote monitoring solution to assess upper limb health by detecting and characterizing GDMs through continuous wrist movement monitoring.