AI reveals fundamental brain differences between women and men

A new deep learning model called “Explainable AI” has uncovered fundamental organizational differences between male and female brains, with more than 90% reliability. The tool stands out for its ability to “explain” the subtle differences it senses, rather than just detect them. These findings may help improve management strategies for neuropsychiatric disorders and illnesses that affect women and men differently.

The hypothesis that sex plays an important role in brain development and function has long been a matter of controversy. However, sex determines, for example, the hormonal flux to which the brain is exposed throughout life, particularly during early development, puberty, and old age.

Previous neuroimaging-based studies have also revealed differences in the brain cytoarchitecture of men and women. It has been reported that male brains are about 10% larger than female brains. On the other hand, certain areas, including the frontal and medial paralimbic, are more extensive in women. These structures are involved in emotion processing, goal setting, motivation, and self-control.

Differences in functional connectivity were also revealed. These data are consistent with the results of recent studies, which indicate gender differences in interoception (physical sensations and the ability to perceive the internal state of the body). This ability – coupled with self-awareness – will be greater in women.

On the other hand, certain neurological and psychiatric disorders and illnesses affect men and women differently. For example, postmenopausal women represent 60% of cases of Alzheimer’s (the most common form of dementia). And according to recent research, brain features associated with motor systems and language differ in autistic males and females.

However, previous studies, focused on how brain regions connect to each other, have not shown clearly consistent, sex-specific brain indices. To establish these differences, a team at Stanford University of Medicine developed a new AI model to perform further analysis.

” One of the main motivations for this study is that gender plays a critical role in human brain development, aging, and the manifestation of psychiatric and neurological disorders. », explains Vinod Menon, lead author of the study, in a Stanford University press release. ” Identifying consistent and reproducible sex differences in the healthy adult brain is a critical step toward a deeper understanding of sex-specific vulnerabilities in psychiatric and neurological disorders. », he adds.

A model explaining the predictions

The model developed by Menon and his team involves a neural network that they call “Explainable AI”. Indeed, models previously used to detect brain differences between the sexes only allowed data to be sorted into two groups, but did not provide an explanation for how this sorting was carried out. In contrast, the new study combs and categorizes large data sets while providing insights into how decisions are made to predict outcomes.

As part of a new study — detailed on the platform
PNAS -, the researchers used their AI model to classify approximately 1,500 dynamic brain MRI images of men and women. This technique makes it possible to capture the complex interactions between different regions of the brain. The data come from various research and health centers located in Europe and the United States.

First, the model identified data belonging to each gender with 90% confidence. Next, he identified the most important brain networks, which allowed him to classify the data. It was found that the AI ​​often turns to the default mode network (activated when you give your thoughts free rein and don’t focus on the outside world), with the striatum and limbic network “deciding” what category to classify the data into. is . The first is involved in the processing of self-referential information, while the second and third play a role in learning and reward processing.

These findings suggest that there are significant differences in the brain by sex, and these have not been reliably identified in previous studies. ” This is very strong evidence that sex is an important determinant of human brain organization », estimates Menon.

Behavioral effects

In the second phase of the study, the team developed other models derived from the first to predict participants’ performance on specific cognitive tasks, based on functional differences linked to sex. The results show that gender-differentiated characteristics have significant behavioral effects.

However, it is important to note that the study could not determine whether these differences appear early in life or whether they may be driven by hormonal differences or changes in the social contexts in which men and women operate.

Still, “this tells me that by ignoring sex differences in brain organization we may be missing key factors that cause neuropsychiatric disorders,” Menon suggests. Additionally, AI models can be applied to study other aspects of brain function. In this regard, the team plans to provide free access to its tool, so that other researchers can take advantage of it.

Source: PNAS