Artificial Intelligence Outcome Prediction in Neonates with Encephalopathy (AI-OPiNE).

Area of Science:

• Pediatric Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Hypoxic-ischemic encephalopathy (HIE) is a major cause of neonatal brain injury.
• Predicting long-term neurodevelopmental outcomes in HIE is crucial for timely intervention.
• Current prediction methods may not fully capture the complexity of HIE-related brain damage.

Purpose of the Study:

• To develop and validate a deep learning algorithm for predicting 2-year neurodevelopmental outcomes in neonates with HIE.
• To utilize multimodal neonatal brain MRI (T1w, T2w, DTI) and basic clinical data for prediction.
• To assess the algorithm's performance on both in-distribution and out-of-distribution test sets.

Main Methods:

• Retrospective analysis of MRI data from term neonates in the HEAL trial (NCT02811263).
• Training of deep learning classifiers (Convolutional Neural Networks) using multisequence MRI and clinical variables (sex, gestational age).
• Evaluation of model performance using area under the receiver operating characteristic curve (AUC) and accuracy on independent test sets.

Main Results:

• The deep learning model achieved an AUC of 0.74 (in-distribution) and 0.77 (out-of-distribution) for predicting death or neurodevelopmental impairment at 2 years.
• Accuracies were 63% (in-distribution) and 78% (out-of-distribution).
• The model showed similar or superior performance for predicting secondary outcomes, including death alone.

Conclusions:

• Deep learning analysis of neonatal brain MRI is a high-performing method for predicting long-term neurodevelopmental outcomes in HIE.
• This AI-driven approach offers a promising tool for pediatric prognosis and guiding clinical management.
• The findings support the use of advanced imaging analysis in neonatal neurocritical care.