A Machine Learning Enhanced Mechanistic Simulation Framework for Functional Deficit Prediction in TBI.

Area of Science:

• Neuroscience
• Biomechanical Engineering
• Computational Modeling

Background:

• Resting state functional magnetic resonance imaging (rsfMRI) offers a non-invasive method to evaluate brain function.
• Assessing functional deficits in brain networks, particularly after traumatic brain injury (TBI), is crucial.
• Current methods for predicting tissue damage from impacts are computationally intensive.

Purpose of the Study:

• To develop a numerical framework for predicting resting state network disruption following head impacts.
• To utilize machine learning for efficient prediction of functional deficits.
• To enable on-the-fly prediction of rsfMRI alterations and reverse-engineered accident reconstruction.

Main Methods:

• A numerical framework combining precalculated impact cases with a machine learning layer was developed.
• The framework predicts resting state network disruption based on head impact parameters (velocity, location, angle, impactor shape).
• Machine learning accuracy was validated using a dummy fall simulation and tested against TBI patient rsfMRI data.

Main Results:

• The machine learning layer demonstrated high accuracy, closely matching full simulation results in a dummy fall case.
• The framework was successfully tested against rsfMRI data from TBI patients, correlating predicted network alterations with injury scenarios.
• The study shows the potential for rapid prediction of rsfMRI alterations from clinical data.

Conclusions:

• The proposed framework offers an efficient method for predicting functional brain network alterations after head impacts.
• This approach facilitates on-the-fly assessment of rsfMRI changes using readily available paramedical data.
• The methodology holds promise for both clinical evaluation and accident reconstruction in TBI cases.