Related Experiment Video
Updated: Jun 3, 2025
10:28
Dynamic Digital Biomarkers of Motor and Cognitive Function in Parkinson's Disease
Published on: July 24, 2019
15.1K
Machine Learning and Statistical Analyses of Sensor Data Reveal Variability Between Repeated Trials in Parkinson's
Rana M Khalil1, Lisa M Shulman2, Ann L Gruber-Baldini3
1Center for Bioinformatics and Computational Biology, University of Maryland, College Park, MD 20742, USA.
Sensors (Basel, Switzerland)
|January 8, 2025
Summary
Repetitive mobility tests for Parkinson's disease (PD) show minimal diagnostic accuracy changes between trials. Variability in subtask duration and sensor data highlights the need for nuanced analysis beyond total task time.
Area of Science:
- Neuroscience
- Biomedical Engineering
- Rehabilitation Science
Background:
- Mobility tasks like Timed Up and Go (TUG) and cognitive TUG (cogTUG) are crucial for assessing Parkinson's disease (PD) impact on motor control, balance, and cognition.
- Wearable sensors offer quantitative measures for evaluating these complex movements.
Purpose of the Study:
- To assess the test-retest reliability of TUG, cogTUG, and walking with turns in individuals with PD and healthy controls.
- To evaluate the performance of machine learning models and statistical metrics using wearable sensor data for PD diagnosis.
Main Methods:
- Collected data from 262 PD participants and 50 controls performing mobility tasks over two trials.
- Utilized wearable-sensor-derived measures, machine learning models, and statistical metrics (including ICC) to analyze total duration, subtask duration, and other quantitative aspects.
- Compared performance between the first and second trials to determine test-retest reliability.
Main Results:
- Diagnostic accuracy for distinguishing PD from controls decreased by only 1.8% between trials, suggesting repetition may not be essential for diagnosis.
- While total task duration showed good consistency (ICC = 0.62-0.95), subtask durations and sensor-derived measures exhibited greater variability between trials.
- This variability differed significantly between controls and PD participants, and across different PD severity groups, underscoring population-specific nuances.
Conclusions:
- Relying solely on total task duration and conventional statistics may oversimplify the reliability of mobility tasks in PD assessment.
- Nuanced analysis of subtask variability and sensor data is essential for accurately capturing movement characteristics in PD.
- Findings highlight the importance of considering population characteristics when interpreting mobility task reliability in PD research and clinical practice.
