Title:
Modelling and Characterization of Force Plate Measurements on Subacute Post-Concussion Subjects Through Machine Learning
Modelling and Characterization of Force Plate Measurements on Subacute Post-Concussion Subjects Through Machine Learning
Author(s)
Casado Garrido, Jose Joaquin
Advisor(s)
Gore, Russell
Editor(s)
Collections
Supplementary to
Permanent Link
Abstract
Mild traumatic brain injuries (mTBI) are one of the leading causes of neurological disorders. Symptoms after a mTBI may include headache, dizziness, and balance issues, among others, with vestibular disorders observed in up to 80% of these patients. These symptoms generally resolve in the first few weeks after the injury, but some patients may develop persistent symptoms. Patients with Post-Concussion Vestibular Dysfunction (PCVD) may present alterations in the peripheral and central vestibular systems. These alterations may then affect postural control and stability, which coupled with visual motion sensitivity, cause the prolonged symptomatology. In this study, we evaluated postural control strategies in Healthy Controls (HC) and Subacute PCVD patients (ST) to identify underlying changes in the postural control system. Sensory Organization Test (SOT) was employed to measure Centre Of Pressure (COP) signals under different sensory conditions. Analysis of traditional linear metrics and entropy metrics of the COP signals demonstrated significant differences between groups. Complexity index was reduced for the ST group during “Eyes Closed” condition, with a median value of 7.93 vs 9.59 for the HC in the Medial-Lateral direction (p=0.002), and 5.17 vs 6.22 Anterior-Posterior direction (p=0.0009). Moreover, analysis of these metrics through machine learning, showed indications of interactions between these variables that may be predictive of the health condition of the patient. These results remark the potential of these metrics for evaluating changes in postural dynamics in patients with PCVD, and opens a new path for analysis of the COP signals with the support of machine learning models.
Sponsor
Date Issued
2021-12-14
Extent
Resource Type
Text
Resource Subtype
Thesis