Related Experiment Video
Updated: Jul 12, 2025
08:56
Clinical Assessment of Spatiotemporal Gait Parameters in Patients and Older Adults
Published on: November 7, 2014
13.9K
Instrumented Gait Classification Using Meaningful Features in Patients with Impaired Coordination
Zeus T Dominguez-Vega1, Mariano Bernaldo de Quiros1, Jan Willem J Elting1
1Department of Neurology, University Medical Center Groningen, University of Groningen, 9713 GZ Groningen, The Netherlands.
Sensors (Basel, Switzerland)
|October 28, 2023
Summary
This study uses gait analysis to differentiate early onset ataxia (EOA) and developmental coordination disorder (DCD) in children. Quantitative gait features achieved higher classification accuracy than clinical assessment, aiding differential diagnosis.
Area of Science:
- Neurology
- Pediatrics
- Biomechanical Engineering
Background:
- Early onset ataxia (EOA) and developmental coordination disorder (DCD) present overlapping symptoms affecting cerebellar function in children.
- Distinguishing between EOA and DCD in pediatric populations is clinically challenging.
- Current diagnostic methods may benefit from objective, quantitative assessments.
Purpose of the Study:
- To develop a machine learning model for classifying children with EOA, DCD, and typically developing controls (CTRL) using quantitative gait data.
- To identify specific gait features that improve diagnostic accuracy and explainability compared to existing approaches.
- To evaluate the performance of a random forest classifier in differentiating these pediatric groups.
Main Methods:
- Collected gait data from 18 EOA, 14 DCD, and 29 CTRL children during Scale for the Assessment and Rating of Ataxia (SARA) gait tests.
- Utilized inertial measurement units (IMUs) to capture movement data and a gait model to extract 36 meaningful features.
- Employed a random forest classifier with leave-one-out cross-validation and synthetic oversampling for classification.
Main Results:
- The random forest model achieved an overall classification accuracy of 82.0%, outperforming clinical assessment (73.0%).
- Mean classification accuracies for EOA, DCD, and CTRL groups were 62.9%, 85.5%, and 94.5%, respectively.
- Key discriminating features included hip flexion-extension range during gait and movement variability during tandem gait.
Conclusions:
- Quantitative gait analysis, particularly features related to hip movement and gait variability, offers a promising tool for the differential diagnosis of EOA and DCD in children.
- Machine learning classification based on SARA-gait data provides a more accurate and explainable approach than traditional clinical assessments.
- This methodology can aid clinicians in distinguishing between these neurodevelopmental conditions.
