# Sex Classification Based on the Functional Connectivity Patterns of the Language Network: A Resting State fMRI Study

## TL;DR

This study uses brain scans to show that resting brain activity in language-related areas can predict a person's sex with high accuracy.

## Contribution

The study introduces a machine learning approach to classify sex based on resting-state functional connectivity patterns in the language network.

## Key findings

- A machine learning classifier predicted sex with 91.3% accuracy using resting-state functional connectivity patterns.
- Key discriminant features were anchored to the left opercular part of the inferior frontal gyrus and other language-related regions.
- Men showed stronger functional connectivity in these regions compared to women.

## Abstract

Research on sex differences in the brain is essential for a better understanding of how the brain develops and ages, and how neurological and psychiatric conditions can impact men and women differently. While numerous studies have focused on sex differences in brain structures, few have examined the characteristics of functional networks, particularly the language network. Although previous research suggests similar overall language performance across sexes, differences may still exist in the brain networks that underlie language processing. In addition, prior studies on sex differences in language have predominantly relied on task‐based fMRI, which may fail to capture subtle differences in underlying functional activity. In this study, we applied a machine learning approach to classify participants' sex based on resting‐state functional connectivity patterns of the language network in healthy young adults (270 men and 288 women; age: 22–36 years), and to identify the most predictive functional connectivity features. The classifier achieved 91.3% accuracy, with key discriminant features anchored to the left opercular part of the inferior frontal gyrus, the left planum temporale, and the left anterior middle temporal gyrus. These regions show distinctive connectivity patterns with heteromodal association cortices, including the occipital poles, angular gyrus, posterior cingulate gyrus, and intraparietal sulcus. Although there was an overlap between men and women, men displayed stronger functional connectivity values in these regions. These findings highlight sex‐related differences in functional connectivity patterns of the language network at rest, underscoring the importance of considering sex as a variable in future research on language and brain function.

We classified the resting‐state functional connectivity patterns anchored to core regions of the language network in a large sample of healthy young adults. The results showed that these patterns contain information capable of predicting sex with an accuracy of 91.3% on the test set and 78.1% on the holdout set.