Prediction of brain maturity in infants using machine-learning algorithms.

Area of Science:

• Neuroimaging
• Developmental Neuroscience
• Machine Learning

Background:

• Large-scale functional brain networks supporting key functions are present in infants.
• Predicting neurodevelopmental outcomes from early brain connectivity is crucial.
• Multivariate pattern analysis is a powerful tool for high-dimensional neuroimaging data.

Purpose of the Study:

• To evaluate resting-state functional MRI data from preterm and term-born infants.
• To determine if machine learning can differentiate between these groups.
• To assess the potential of these methods for predicting neurodevelopmental outcomes.

Main Methods:

• Resting-state functional MRI scans from 50 preterm and 50 term-born infants at term equivalent age.
• Analysis using 214 regions of interest and binary support vector machines.
• Application of support vector regression for quantitative estimation of gestational age.

Main Results:

• Support vector machines accurately distinguished preterm from term infants (84% accuracy).
• Both inter- and intra-hemispheric connections were key in group categorization.
• Support vector regression quantitatively estimated birth gestational age from resting-state functional MRI data.

Conclusions:

• Preterm birth is associated with detectable widespread changes in brain functional network architecture by term equivalent age.
• Machine learning analysis of resting-state functional MRI can identify these early alterations.
• This approach shows promise for predicting neurodevelopmental outcomes in individual preterm infants.