Machine-learning algorithms can now estimate the "brain age" of infants with unprecedented precision by analyzing electrical brain signals recorded using electroencephalography (EEG).
A team led by Sarah Lippé at Université de Montréal's Department of Psychology has developed a method that can determine in minutes whether a baby's brain development is advanced, delayed or in line with their chronological age.
This breakthrough promises to enable early screening and personalized monitoring of developmental disorders in babies.
"The first years of life are critical for brain development," explained Lippé. "This is when changes at all levels of brain structure and function lay the foundation for increasingly complex information processing."
This fact underscores the importance of reliable tools for the assessment of brain maturation and early identification of children at risk for neurodevelopmental disorders, such as language delays, attention deficit disorder and autism.
According to Lippé, early intervention and close monitoring can significantly improve long-term outcomes.
272 babies studied
The study examined 272 babies, 53 of whom had macrocephaly, a condition characterized by an abnormally large head and associated with atypical brain development.
Under Lippé's supervision, PhD student Saeideh Davoudi compared two methods for analyzing the babies' EEGs: conventional machine learning and novel deep learning.
To train the conventional machine-learning model, Davoudi extracted key features from the EEGs, including signal complexity and the intensity of brain wave activity in the delta, theta and alpha frequency ranges.
For the novel deep-learning model, Davoudi fed the raw EEG data directly into the model, which automatically analyzed the data for patterns.
The results, published in May 2025 in the journal NeuroImage, showed that the novel deep-learning model performed best.
"From just a few minutes of EEG signal, we were able to estimate a baby's age with a mean error of less than 30 days," reported Lippé. "This is a powerful tool for detecting delays and accelerations in brain maturation."
The study highlights how brain waves are a key marker of brain age. For example, alpha waves (6-9 hertz), which are associated with attention and relaxation, become more pronounced as the baby develops, reflecting its growing integration of cognitive functions.
Conversely, delta waves (0.5-2.5 hertz), which are characteristic of deep sleep, predominate in babies but become less frequent as the brain matures.
Can detect anomalies
In addition to estimating brain age, this non-invasive tool can detect anomalies in the pace of neurodevelopment. For example, the study found that babies with macrocephaly exhibited delays in brain maturation compared to those without the condition.
It also showed that brain age was correlated with measures of behavioural and cognitive function, as assessed by tests of adaptive behaviour and information-processing speed.
According to Lippé, these findings pave the way for new clinical applications.
"Estimated brain age could help us identify children at risk for developmental disorders before behavioural symptoms appear," she said. "It could also be used to monitor the effectiveness of therapeutic interventions by providing an objective indicator of how brain development is progressing."
About this study
"Electroencephalography estimates brain age in infants with high precision: Leveraging advanced machine learning in healthcare," by Saeideh Davoudi, Sarah Lippé et al., was published May 15, 2025 in NeuroImage.
