In civilian populations, it is estimated that approximately 1.6-3.8 million concussions occur each year. In military populations, 380,000 traumatic brain injuries have occurred from 2000-2018. The most commonly reported pain-related symptom following concussion is acute post-traumatic headache (PTH). Although some people recover from PTH during the acute phase, a substantial number of concussed patients with PTH develop persistent post-traumatic headache (PPTH), which is associated with high-risk of prolonged disability. There is currently no prognostic biomarker for predicting whether a concussed patient will recover from acute PTH or will develop PPTH. This is an unmet need and a dilemma for making early treatment decisions, for prognosticating recovery and for developing non-opioid treatment therapies.
We will leverage our prior experience in building machine learning models to develop a prognostic biomarker signature for PPTH based on neuroimaging and clinical data collected shortly after concussion. Furthermore, to optimize reproducibility and validation of the biomarker signature, two machine-learning laboratories will build and validate classification models. We hypothesize that this biomarker signature will have high accuracy for prognosticating which patients with acute PTH are at high risk of developing PPTH.