Pull Test estimation in Parkinson's disease patients using wearable sensor technology.

Area of Science:

• Neurology
• Biomedical Engineering
• Movement Disorders

Background:

• Postural instability is a key motor impairment in Parkinson's disease (PD).
• The clinical Pull Test is widely used but suffers from subjectivity and low discriminative power.
• Objective and reliable methods are needed to quantify postural instability in PD.

Purpose of the Study:

• To develop a novel methodology for estimating Pull Test scores in PD patients.
• To explore the relationship between foot motion patterns and Pull Test outcomes.
• To identify optimal motor function tests for predicting postural instability based on PD phenotype.

Main Methods:

• 139 idiopathic PD patients underwent motor function tests while wearing shoe-mounted inertial sensors.
• Acceleration and orientation data were recorded, from which 684 features were extracted.
• Feature selection and classification algorithms were employed to estimate Pull Test scores and identify predictive tests.

Main Results:

• A classification accuracy of 0.75 (CI: [0.69-0.82]) was achieved when combining all phenotypes and tests.
• Foot circling best predicted scores for equivalent (0.79) and bradykinetic (0.75) PD phenotypes.
• The 2x10m walk with stop-and-go was superior for tremor-dominant patients (0.75).

Conclusions:

• Inertial data from foot motion can reliably estimate postural instability in Parkinson's disease.
• This sensor-based approach offers an objective alternative to the subjective Pull Test.
• Specific motor tasks show differential predictive power for postural instability across PD phenotypes.