Wearable Movement-Tracking for Prodromal Parkinson's Disease Detection: A Cross-Country Validation Study

Area of Science:

• Neurology
• Biomedical Engineering
• Data Science

Background:

• Accelerometer-based models show promise for detecting prodromal Parkinson's disease (PD).
• Generalizability of models trained on UK Biobank (UKBB) data is uncertain due to diagnostic timing inconsistencies.
• This study addresses the need for external validation of PD detection models.

Purpose of the Study:

• To evaluate the performance of existing prodromal PD detection models in international cohorts.
• To assess the generalizability of models trained on UK Biobank data.
• To compare model performance across different populations with rapid eye movement sleep behavior disorder.

Main Methods:

• Applied previously published models to German and British cohorts.
• Included individuals with isolated/idiopathic rapid eye movement sleep behavior disorder and healthy controls.
• Compared hourly acceleration patterns and classification performance across cohorts.

Main Results:

• British cohort showed similar patterns to UKBB but weaker statistical differences and reduced performance.
• German cohort exhibited no significant group differences and lower performance.
• No cohort pairs demonstrated statistical equivalence, indicating poor generalizability.

Conclusions:

• UK Biobank models may identify early clinical PD, not universal prodromal signs.
• Prospective validation in well-characterized, diverse cohorts is essential.
• Current models require further refinement for reliable clinical translation.