Objective Classification of mTBI Using Machine Learning on a Combination of Frontopolar Electroencephalography...

Area of Science:

• Neuroscience
• Medical Technology
• Sports Medicine

Background:

• Mild traumatic brain injury (mTBI) diagnosis remains a significant challenge in sports and military settings.
• The cumulative effects of head injuries can negatively impact long-term cognitive function and quality of life.
• Despite advancements, individuals often face pressure to continue activity despite mTBI symptoms.

Purpose of the Study:

• To evaluate the accuracy of a novel system for immediate, objective mTBI assessment.
• To determine if electroencephalography (EEG) can enhance sideline mTBI detection.
• To develop a tool for on-field or in-locker-room mTBI evaluation.

Main Methods:

• A cohort of 38 individuals with recent mTBI and 47 healthy controls participated.
• Resting-state EEG was recorded using frontopolar electrodes, alongside a symptom questionnaire and behavioral tests.
• A machine learning algorithm (boosting) classified participants using EEG power spectral densities and questionnaire data.

Main Results:

• The combined approach of EEG and symptom questionnaire achieved a diagnostic accuracy of approximately 91% ± 2%.
• The symptom questionnaire alone yielded an accuracy of 82% ± 4%.
• Leave-one-out cross-validation confirmed the significant improvement in accuracy with EEG integration.

Conclusions:

• EEG measurements show significant potential for improving the diagnosis of suspected brain injuries.
• This study represents a step towards objective, field-deployable mTBI assessment tools.
• Integrating EEG offers a more reliable method for classifying brain injury status compared to symptom-based assessments alone.