To utilize machine learning models to classify recovery duration for medical clearance (return to sport) following sport-related TBI and evaluate the predictive accuracy of longitudinal data.
Key Findings:
Machine learning can improve predictive accuracy for time to clearance from sport-related concussions, potentially transforming clinical practice.
Longitudinal data enhances prediction capabilities compared to single time point assessments, indicating a need for comprehensive data collection.
Identifying specific assessment features can inform evidence-based protocols for concussion management, leading to better patient outcomes.
Interpretation:
The study supports the potential of machine learning to enhance clinical decision-making regarding return-to-play timelines for athletes with concussions, emphasizing the need for data-driven approaches.
Limitations:
External validation and prospective testing are required before clinical deployment, highlighting the need for rigorous testing of ML models.
The study does not propose a single deployable clinical tool but benchmarks multiple classifiers, indicating the complexity of developing a validated prediction tool.
Conclusion:
The findings provide a framework for longitudinal monitoring of TBI, aiding individualized return-to-play decisions and informing broader clinical guidelines.
