Each year in the US there are approximately 2.8 million Traumatic Brain Injury (TBI)-related ED visits, hospitalizations, and deaths, and over 2 million individuals are diagnosed with mild TBI or concussion. The personal and societal costs are enormous. Post-traumatic headache (PTH) is the most common symptom following concussion or TBI. Although some PTHs resolve within the first few days, for many with PTH the headaches last for months or years. We have been involved in a number of projects that develop machine learning algorithms to study the mechanisms, build classifiers, and identify prognostic signatures based on a combination of neuroimaging, clinical phenotyping and neurophysiology, and molecular biomarker and genetic datasets within the context of TBI, concussion, PTH and migraine.
Methodologies:
- Multi-modality data fusion and predictive models
- Hierarchical modality and feature selection in sparse learning
- Spatial-temporal models for neuroimages
- Functional data analysis
- Reproducibility analysis