Summary:
UCLA researchers have developed a machine learning-based clinical decision support tool that predicts patient-specific cognitive impairments following traumatic brain injury (TBI) using neuroimaging and multimodal clinical data. The software leverages MRI-derived structural brain information to generate individualized diagnostic and prognostic insights, enabling clinicians to anticipate outcomes such as depression, PTSD, and cognitive decline within months post-injury.
Background:
Traumatic brain injury (TBI) is a leading cause of long-term neurological disability, affecting over 5.3 million individuals in the United States and millions more globally each year. TBIs range in severity from mild concussions to severe brain damage, often resulting in persistent cognitive, emotional, and behavioral impairments. These may include deficits in memory, attention, executive function, mood regulation, and increased susceptibility to conditions such as depression and post-traumatic stress disorder (PTSD).
Despite the high prevalence and heterogeneity of TBI outcomes, current clinical workflows lack precise tools to predict individualized cognitive trajectories. Standard treatment protocols often apply generalized care pathways, such as limited follow-up or uniform discharge recommendations, which fail to account for patient-specific brain structure, injury patterns, and resilience variability. This results in suboptimal allocation of medical resources and missed opportunities for early intervention in high-risk patients. There is a critical unmet need for tools that can bridge the gap between observed neural damage and downstream cognitive outcomes to support personalized care and optimized patient outcome.
Innovation:
UCLA researchers have developed a novel, software-based platform that integrates neuroimaging and machine learning to directly link structural brain damage with predicted cognitive impairments on a per-patient basis. The tool uses MRI data as its primary input and applies advanced predictive modeling to forecast specific neuropsychiatric and cognitive outcomes, including depression, PTSD, and cognitive dysfunction, up to three months post-injury. This approach is grounded in a first-of-its-kind scientific framework that quantitatively correlates neural injury patterns with cognitive and behavioral outcomes. The model captures inter-patient variability, demonstrating that individuals with similar injuries and medical histories can experience significantly different cognitive trajectories—ranging from high resilience to severe impairment.
In addition to MRI data, the platform is designed to incorporate multimodal inputs such as blood-based biomarkers and cognitive assessment scores (e.g., questionnaires), enhancing predictive accuracy and clinical flexibility. The system is adaptable to a wide range of neurological and psychological conditions beyond mild TBI, including moderate to severe (e.g., coma patients), and conditions such as Alzheimer’s Disease and stroke. By automating the interpretation of complex neuroimaging data and translating it into clinically actionable predictions, this tool enables a shift toward precision medicine in neurotrauma care. It supports early identification of high-risk patients and facilitates targeted interventions, including referrals, monitoring, and treatment planning.
Potential Applications:
● Clinical management of traumatic brain injury based on severity
● Mild TBI (concussion)
● Moderate to severe (e.g., coma patients)
● Military and veteran healthcare (e.g., PTSD and blast injuries)
● Sports medicine and concussion management (return-to-play protocols)
● Neurology and psychiatry decision support systems
● Rehabilitation planning and cognitive therapy allocation
● Extension to other neurological and psychiatric disorders
● Alzheimer’s Disease
● Stroke
Advantages:
● Patient-specific prediction of cognitive outcomes based on brain structure
● Early prognostic capability (up to 3 months post-injury)
● Integration of multimodal data (MRI, biomarkers, cognitive assessments)
● Enables personalized treatment and resource allocation
● Identifies high-risk patients who may otherwise be overlooked
● Flexible framework adaptable to multiple neurological conditions
● Improves clinical decision-making and care efficiency
State of Development:
First description of the complete invention June 2025. Prototype software developed and validated using clinical datasets, including mild TBI patient cohorts. Model development and validation are documented in dissertation research, with demonstrated predictive capability linking neural damage to cognitive outcomes.
Related Publications and Patents:
Details available upon request. Foundational work described in associated dissertation chapters investigating neural-cognitive relationships and predictive modeling.
Reference:
UCLA Case No. 2026-251
Inventors:
Sonya Ashikyan, Martin Monti, Jeffrey Chiang