Explainable Machine Learning for Coma Outcome Prediction Based on Structural and Functional Brain MRI

Key Points

• The aim is to predict coma patient outcomes based on brain MRI data, assessing recovery potential.
• Developed an ensemble of metrics derived from brain MRI.
• Focused on structural and functional analysis of brain networks.
• Evaluated correlations with coma state and patient etiology.
• Identified key MRI-derived metrics associated with coma state and recovery.
• Mesocircuit and frontoparietal networks were found to provide vital information.
• Predictions significantly correlated with patient recovery at 3 months.