Yale Researchers Use Machine Learning To Identify Brain Networks Predictive Of Aggression In Children

Source: https://news.yale.edu/2021/10/26/machine-learning-reveals-brain-networks-involved-child-aggression

Children’s mental disorders are defined as significant changes in how children learn, behave, or handle their emotions. Many times these changes create discomfort and make it difficult for them to get through the day. Outbursts of anger and physical hostility are common in children with psychiatric problems such as attention deficit hyperactivity disorder (ADHD). A deeper understanding of what causes these symptoms could aid in the development of therapy options. Many researchers and healthcare practitioners have been looking to use machine learning to diagnose mental health issues in children.

A study conducted by Yale researchers has, for the first time, discovered changes in brain connectivity in children who demonstrate aggression using a machine learning-based technique.

Unlike earlier studies that focused on individual brain regions, the current research identifies patterns of neural connections linked to aggressive behavior in children across the entire brain. 

The study is based on a new model of brain activity known as the “connectome,” which defines the pattern of brain-wide connections. Maladaptive aggression can be harmful to oneself or others. One of the most common reasons for referrals to child mental health services is this challenging behavior. Connectome-based modeling provides a new understanding of the brain networks involved in aggressive behavior.

Initially, researchers collected fMRI (functional magnetic resonance imaging) data. For this, the children were given an emotional face recognition task in which they viewed faces giving different expressions such as calm, scared, etc. According to the researchers, seeing faces that convey emotion can activate brain states related to emotion creation and control. Both of these brain states are associated with aggressive behavior. They next used machine learning to uncover brain connections that differentiated children with and without aggressive behavior histories.

The results predicted the patterns in brain networks engaged in social and emotional activities, such as feeling annoyed with homework or understanding why a friend is sad. The predictions suggest that these patterns are related to aggressive behavior. 

Further, the researchers tested their idea on subgroups of children with aggressive behavior and disorders like anxiety, ADHD, and autism. The results imply that abnormal connectivity to the dorsolateral prefrontal cortex is crucial in regulating emotions and higher cognitive functions, including attention and decision-making. According to researchers, this abnormal connectivity has emerged as a consistent predictor of aggression.

The researchers suggest that future clinical studies can use the new robust large-scale brain networks and their connectivity with the prefrontal cortex to represent a neural marker of aggression. The extensive interconnection of the brain is described by the human functional connectome. Understanding the connectome is at the cutting edge of neuroscience since it can provide crucial information for building psychiatric biomarkers.

The team believes that the connectome model of aggression can be used to design treatment interventions to improve the synchronization of these brain networks could aid future studies.

Research Paper: https://www.nature.com/articles/s41380-021-01317-5

Source: https://news.yale.edu/2021/10/26/machine-learning-reveals-brain-networks-involved-child-aggression