AI Model Integrating Imaging and Clinical Data for Predicting CSF Diversion in Neonatal Hydrocephalus: A Preliminary...

Area of Science:

• Neurology
• Medical Imaging
• Artificial Intelligence

Background:

• Neonatal hydrocephalus requires timely cerebrospinal fluid (CSF) diversion.
• Current predictive tools for stratifying hydrocephalus risk are lacking.
• Accurate prediction of CSF diversion needs is crucial for optimal neonatal care.

Purpose of the Study:

• To develop and validate an artificial intelligence (AI) model for predicting CSF diversion needs in neonates.
• To integrate multimodal imaging and clinical data for enhanced predictive accuracy.
• To stratify neonatal hydrocephalus into low- and high-risk groups.

Main Methods:

• Development and validation of a hybrid AI model using multimodal data (MRI and clinical information).
• Inclusion of development (n=116) and external validation (n=21) cohorts of neonates.
• Performance assessment using Area Under the Receiver Operating Characteristics Curve (AUC), sensitivity, and specificity.

Main Results:

• The AI model achieved an AUC of 0.824 in the development cohort and 0.808 in the external validation cohort.
• The hybrid AI model significantly outperformed clinical-only and image-only models in predicting raised intracranial pressure.
• Accurate prediction of CSF diversion was observed across various etiologies, including post-hemorrhagic hydrocephalus and myelomeningocele closure.

Conclusions:

• The developed AI model demonstrates robust performance in predicting the need for CSF diversion in neonates.
• The AI model can assist clinical decision-making, especially in resource-limited settings.
• Further refinement is needed for complex etiologies like myelomeningocele-associated hydrocephalus.