Once college athletes are diagnosed with a concussion caused by a blow to the head, neck or body that affects the brain, they want to know when they will recover so they can return to playing sports again. While the typical recovery time for college athletes with concussions is within four weeks, about 15 to 20 percent of them take longer to recover because of persistent post-concussion symptoms (PPCS). These symptoms include headaches, dizziness, irritability, and loss of concentration and memory.
A prognostic model that could predict the risk of late recovery with greater accuracy than current clinical models that rely on symptom data would help clinicians start treatment and other interventions earlier.
A collaborative team of NIH-funded researchers developed a new approach based on their analysis of an advanced type of brain imaging data and return-to-play clinical records. The results of their study published in Neuroimage: Clinical found that imaging data related to specific white matter brain regions had a significant correlation with return to play and that their model could correctly identify early versus late recovery 90% of the time.
“This model could improve the prognostic accuracy of assessing an individual’s outlook following a concussion. These advances, including the new automated platform, will serve as part of the infrastructure supporting neuroscience research and could lead to new discoveries,” said Qi Duan, Ph.D., a program director in NIBIB’s Division of Health Information Technology.
How they built the concussion recovery model
The researchers wanted to determine whether differences in microstructural properties of the brain’s white matter were associated with return to play. Previous studies have shown that concussions typically affect white matter, a deep brain region that facilitates communication between different areas of the brain.
Their analysis focused on return-to-play records and an advanced type of magnetic resonance imaging (diffusion MRI), which allows the movement of water molecules in tissues to be measured, providing detailed images of the brain.
“Following a concussion, both patients and families are really affected by the patient’s inability to return to normal activities. Currently, clinicians lack the tools to predict when a patient with a concussion can return to work and resume normal activities. To address that gap in clinical practice, we developed a tool that uses machine-learning and diffusion MRI, a widely available technology, to predict when patients with concussions can return to normal life activities,” said Franco Pestilli, Ph.D., co-corresponding author and professor of psychology and neuroscience at the University of Texas at Austin.
The researchers analyzed the diffusion MRI (dMRI) data for 51 injuries of 45 student athletes that was a subset of the National Collegiate Athletic Association and Department of Defense Concussion Assessment, Research, and Education Consortium (CARE) study. They designed an automated cloud computing platform with NIBIB and NIMH funding to process and analyze the neuroscience data. The platform (brainlife.io) is available to other neuroscience researchers with dMRI data obtained by scanning various patient populations with concussions.
The return-to-play data was divided into two groups: athletes who returned in less than 28 days and athletes who returned in 28 days or more. The late return-to-play group met the definition of PPCS.
The researchers developed a data analysis approach that used the automated platform to extract the microstructural properties of different white matter brain structures. For each region, they calculated two established measures for the movement of water molecules in white brain tissue: fractional anisotropy (FA), which indicates how freely water molecules can move in multiple directions, and mean diffusivity (MD), which measures the rate of water movement.

The researchers found that only the FA measure of all white matter regions correlated significantly with the return-to-play data.
Subsequent analyses focused on whether FA data for specific white matter tracts (bundles of fiber that connect different parts of the brain), could predict early versus late recovery for athletes.
Focusing on the 16 tracts with the strongest data, the researchers developed a model to predict which athletes would take longer than 28 days to recover. Using this model, the researchers could correctly categorize the athletes 90% of the time.
“Overall, this preliminary study demonstrates the possibility of using the microstructural properties of white matter tracts to develop a prognostic model for PPCS, which outperforms current predictive models based on clinical data,” said co-corresponding author Nicholas Port, Ph.D., professor of optometry and adjunct professor of psychological and brain sciences at Indiana University in Bloomington.
These findings describe preliminary research. A significant limitation was the small number of injuries (51) included in the study, which highlights the need for larger dMRI concussion studies focused on recovery. Another limitation is that the subjects were all college athletes.
The researchers plan to address those limitations in a large clinical trial involving a broader set of patients with concussions.
“Once we have collected the data, we would like to use it to train the machine learning model again to make it more robust and ultimately clinically impactful,” said Pestilli.
This study, in particular the cloud computing platform (brainlife.io), was supported in part by NIBIB grants R01EB030896 and R01EB029272, and grants from National Institute of Mental Health (R01MH126699, R01MH13370), National Institute of Neurological Disorders and Stroke (UM1NS132207, U24NS140384) and the National Science Foundation (OAC-1916518, IIS-1912270, IIS-1636893, and BCS-1734853).
Additional research was supported by the Department of Defense (W81XWH-20-1-0717, W81XWH-14-2-0151).
This science highlight describes a basic research finding. Basic research increases our understanding of human behavior and biology, which is foundational to advancing new and better ways to prevent, diagnose, and treat disease. Science is an unpredictable and incremental process—each research advance builds on past discoveries, often in unexpected ways. Most clinical advances would not be possible without the knowledge of fundamental basic research.
Study reference: G Berto et al. Diffusion tensor analysis of white matter tracts is prognostic of persisting post-concussion symptoms in collegiate athletes. Neuroimaging: Clin. 2024. Doi: 10.1016/j.nicl.2024.103646.