Classification of neuromechanical control strategy in a wrist rotation task

Ziegelman, Liran

Permalink

https://hdl.handle.net/2142/124314 Copy

Description

Title

Classification of neuromechanical control strategy in a wrist rotation task

Author(s)

Ziegelman, Liran

Issue Date

2024-04-19

Director of Research (if dissertation) or Advisor (if thesis)

Hernandez, Manuel

Doctoral Committee Chair(s)

Hernandez, Manuel

Committee Member(s)

• Hsiao-Wecksler, Elizabeth
• Sowers, Richard
• Koyejo, Oluwasanmi

Department of Study

Neuroscience Program

Discipline

Neuroscience

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• aging
• classification
• control strategy
• EEG
• machine learning
• motion
• motor
• neuromechanics
• NODE

Abstract

Speed-accuracy trade-offs exist in a variety of neural tasks. It is the goal of this work to establish the presence of and predictability of a motor control strategy during a wrist rotation task. Participants were asked to perform a series of continuous and discrete wrist rotations. This motion data was clustered into segments of either speed or range of motion oriented control strategy, controlling for age cohort, continuity of task, and motion type. Age-related changes in motion and cortical data were explored, as were control strategy related changes. Finally, competing neural ordinary differential equation (NODE) and random forest models were fit to explore the ability to classify control strategy using cortical data alone. The clustering method was found to be successful in establishing control strategy. While age-related changes were not prevalent in direct exploration of motion data, older adults are found to have a lower speed with prioritizing speed as compared to young adults doing the same. Control strategy differences were present in the primary motor cortex at the N1, N2, and P3 components, at the supplementary motor area in the P1 component, and in the prefrontal cortex using prefrontal negativity. When using both motor and prefrontal cortical inputs to competing models, models perform with a similar accuracy but the NODE model was able to train and test data at a much faster pace compared to the random forest.

Graduation Semester

2024-05

Type of Resource

Thesis

Copyright and License Information

Copyright 2024 Liran Ziegelman

Owning Collections