Unveiling Brain Sex Differences with AI

Stanford Medicine researchers have developed an artificial intelligence model that can distinguish between a woman’s and a man’s brain-activity scans with better than 90% accuracy. The unexpected result, reported Feb. 20 in the Proceedings of the National Academy of Sciences puts to rest at last a longstanding argument over whether there are real sex differences in the human brain. It also outlines how critical such differences are in the treatment of neuropsychiatric disorders, which might differentially affect men and women.

One crucial aspect that the senior author, Vinod Menon, PhD, points out is that sex has an enormous influence on brain development, aging, and neurological and psychiatric disease manifestations. This new research enables Menon, with lead authors Yuan Zhang, PhD, and Srikanth Ryali, PhD, to make a stride toward defining just what has been an essential step toward understanding sex-specific vulnerabilities in neurological and mental diseases: persistent and reproducible sex differences in the healthy adult brain.

Such success of the AI model is because it analyzed the dynamic MRI scans that are a representation of complex relationships between different parts of the brain. The “hotspots” that gave the most significant contribution to helping the model tell the male and female brains apart were the striatum and limbic network participating in learning and reward response, and the default mode network which is involved in processing self-referential information.

The paper does not take a stand on whether these sex-related differences are due to hormonal differences and social situations faced by men and women or whether they are inborn from an early age. Whatever the reason, however, the difference exists, and the high accuracy of the AI model in identifying them speaks volumes about how good sex is as a predictor of human brain structure.

‘Explainable AI’ was employed by the researchers to pinpoint the brain networks that played a crucial role in the model’s judgments. This technology enables a more thorough comprehension of the AI model’s decision-making process, which in this instance indicated the limbic, striatal, and default mode networks.

The team also investigated the potential for using sex-specific functional brain characteristics to predict cognitive task performance. In order to demonstrate that functional brain differences between the sexes had important behavioral ramifications, they created sex-specific models that accurately predicted cognitive ability.

The study’s broad ramifications include improving our understanding of the brain and creating new opportunities for tailored medicine by identifying sex differences. The study, which has NIH support, will be published in PNAS. For wider research applications, Menon and his team will make their AI model available.

Researchers from Stanford Medicine have made a major advancement in neuroscience with this study, which provides strong proof of sex variations in brain structure and pave the way for more individualized medical care.