Research Report
Population: Adult
Karen L McCulloch, PT, PhD, MS, NCS, Professor, UNC-Chapel Hill kmac@med.unc.edu
Oleg Favorov, PhD, Research Associate Professor, UNC-Chapel Hill, Biomedical Engineering olegfavorov@gmail.com
Keywords: Prediction, Mild Brain Injury, Concussion, Accelerometry
Introduction: Recent advances in medical science have highlighted the need for methods to quantify the effects of mild traumatic brain injury (mTBI). The ability to determine when an individual has recovered from injury is a particular challenge. Typical approaches use neurological examination, neuropsychological tests, standing balance tests or self-report of post-concussive symptoms. If impairments are subtle, these methods may not detect impairments that could have significant implications for military duty. Quantitative evaluation of performance on motor tests with accelerometry can be both practical and objective, with the potential to be highly sensitive to even subtle neural abnormalities associated with mTBI. This approach has been used with some success in a stand and walk test post-concussion, but performance on higher level mobility tasks (run, obstacle avoidance, etc) have a stronger relevance for military training and combat related activities.
Purpose: We investigated whether the Illinois Agility Test (IAT) might be sensitive to subtle mTBI impairments. The IAT requires a transition from prone to standing, running short distances while navigating multiple obstacles, changing direction, and speed accelerations/decelerations, placing high demands on the sensorimotor system.
Subjects: Recruited from active duty Army service members aged 18-42 stationed at Ft Bragg, North Carolina. Our preliminary sample consisted of 19 subjects diagnosed with mTBI who were undergoing comprehensive outpatient rehabilitation at Womack Army Medical Center, Department of Brain Injury Medicine. All mTBI subjects were at least 30 days post injury, with persistent symptoms preventing return to duty (RTD). 25 healthy control subjects were recruited following a TBI briefing required of all soldiers who transfer to Ft Bragg.
Methods: Two tri-axial accelerometers (1 on an adjustable headband, 1 on a belt at lumbar area) were worn during the test, allowing for continuous wireless transmission of time series data by each of the 6 sensors during the IAT. Following a ‘walk-through’ of the IAT and a practice trial to ensure the task was understood, each subject ran a test trial at full speed. The power spectrum of each subject’s time series for this trial was computed for each accelerometer using Fast Fourier transform. Power spectra of all subjects were averaged to identify the dominant frequencies. The top 35 frequencies were used as the input to a linear Support Vector Machine (SVM). The SVM was trained to distinguish between the control and mTBI subjects. Cross-validated using the standard leave-one-out approach, the SVM was found to correctly classify 23 out of 25 control subjects (92% accuracy) and 10 out of 19 mTBI subjects (52.5% accuracy) based on anterior-posterior torso accelerometer readings. The other 5 sensors were less accurate.
Conclusions and clinical relevance: In a group of subjects with chronic mTBI, performance on the IAT using non-linear methods provided a high level of accuracy for identifying healthy control subjects. The mTBI group included some individuals who were nearing readiness for RTD as a result of rehabilitation, which may explain lower accuracy in identifying those with mTBI. Our group plans to optimize this method on a larger military sample that must perform at a high level of physical function under conditions of significant risk. The ability to identify subtle movement abnormalities during tactical training activities stands to aid therapists in predicting readiness for RTD in active duty military populations.
Citation:
McCulloch, Karen L, PT, PhD, MS, NCS; Favorov, Oleg V, PhD. Agility Task Performance After Military Mild Traumatic Brain Injury: An Approach to Identify Subtle Motor Control Impairments.. Poster Presentation. IV STEP Conference, American Physical Therapy Association, Columbus, OH, July 17, 2016. Online. https://u.osu.edu/ivstep/poster/abstracts/086_mcculloch-favorov/