From brain to education through machine learning: Predicting literacy and numeracy skills from neuroimaging data

Area of Science:

• Educational neuroscience
• Neuroimaging
• Machine learning

Background:

• Educational neuroscience explores using neural data to predict academic success.
• Machine learning (ML) and neuroimaging advancements offer new potential in this field.
• Previous research has explored predicting literacy and numeracy outcomes.

Purpose of the Study:

• To review neuroimaging studies using ML to predict literacy and numeracy.
• To analyze methodologies, including cross-sectional and longitudinal designs, and ML approaches.
• To identify challenges and future directions in the field.

Main Methods:

• Systematic review of neuroimaging studies employing machine learning.
• Analysis of studies predicting literacy and numeracy in diverse populations (children, adults, learning disabilities, typical performance).
• Examination of cross-sectional and longitudinal designs, regression, and classification techniques.

Main Results:

• ML in neuroimaging shows potential for predicting literacy and numeracy outcomes.
• Significant variability exists in algorithms, brain circuits, and methodologies across studies.
• A notable gap exists in longitudinal prediction studies for young children before formal education.

Conclusions:

• ML-enhanced neuroimaging holds promise for educational neuroscience applications.
• Standardization of methods and increased use of portable neuroimaging are crucial for progress.
• Future research should focus on longitudinal prediction in early childhood and accessible neuroimaging.