Miolane – AI Model of the Maternal Brain
ECE Ass't. Prof. Nina Miolane seeks to provide new insights into how the brain changes during pregnancy and motherhood

From the COE News article – "Leveraging AI to Study Women’s Brains During and After Pregnancy"
A woman’s brain undergoes a great deal of change during and after pregnancy, and while one possible result of such changes — postpartum depression — affects some ten to twenty percent of mothers around the world, that condition, and the state of the ever-changing brain of expectant and new mothers, remains a seriously understudied subject. Nina Miolane, an assistant professor in the Department of Electrical & Computer Engineering (ECE) at UC Santa Barbara, recently received a $1 million grant from the Chan Zuckerberg Initiative (CZI) to pursue a project titled “An AI Model of the Maternal Brain.” Miolane seeks to provide new insights into how the brain changes during pregnancy and motherhood and how such changes might shed light on postpartum depression and other issues.
“We’re incredibly honored to receive this support,” said Miolane, who directs the Geometric Intelligence Lab at UCSB and is co-director of AI for the nation-wide Ann S. Bowers Women's Brain Health Initiative (WBHI), headquartered at UCSB. “The maternal brain is an important but overlooked area of science. This project is aimed at bringing it the attention and rigor it deserves.”
The WBHI website reads, “Most of what we know about health and disease centers on the male body. Neuroscientists overlook aspects of the human condition relevant to half of the global population (e.g. the menstrual cycle, the pill, pregnancy, and menopause). The Ann S. Bowers WBHI advances the study of women's brain health through deeply collaborative science.”
At the heart of the project lies what is referred to as the Bowers WBHI Maternal Brain Project (MBP) dataset, a first-of-its-kind resource for mapping neuroanatomical changes — from preconception through two years postpartum — for a diverse population of first-time mothers in the U.S. and Spain.
“With its multimodal data on cognitive assessments, reproductive health, and proteomics [the study of a complete set of proteins expressed by an organism or a cell at a given time], the MBP provides the foundation for our transformative AI effort,” reads the project abstract. “Together, these data-driven insights and AI advancements are aimed at bettering our understanding of maternal mental health, an urgent but historically under-researched public-health priority,” the examination of which, adds Miolane, “is crucial to alleviating the suffering of millions of women every year.”
In the project, Miolane explains, “Essentially, we use AI to predict the future state of the brain, down to its cortical and subcortical structures, and validate these predictions against empirical measurements.”
To accomplish that, she and her team are combining their geometric expertise with AI to create dynamic 3D models — also called atlases or, more commonly, templates — of the maternal brain, which changes based on hormone levels, physiology, and even demographics. The model is connected to an interactive web app that allows a user to adjust sliders on the screen to change variables such as gestational week or hormone levels and see how the change is reflected in the brain’s structure and function. “These models will identify patterns of brain evolution, uncover population-level differences, and predict postpartum conditions such as depression,” Miolane says.
The project is based both at UCSB and in Spain, where a professor named Susanna Carmona and a researcher, Magdalena Martinez-Garcia, are assembling a huge data set containing hundreds of images of women's brains, as well as responses to a questionnaire intended to evaluate depression, making it possible to correlate specific hormone levels with the likelihood of postpartum depression. “We have two data sets,” Miolane says. “The first has brain images of ten women [including Miolane, who gave birth to a son in September 2024], who were sampled every two weeks through pregnancy and motherhood in the lab of Emily Jacobs, associate professor in the UCSB Department of Psychological & Brain Sciences. The women in the second data set were sampled only twice, but the data includes hundreds of participants. By combining the two data sets — one that has a lot of time points and the other that has a lot of women — we can build a complete model.”
One challenge of the project is that the researchers need to design specific AI methods, because, Miolane notes, “While we have the largest data sets on the topic, an AI method usually requires millions of images, and we don’t have that in neuroimaging for women’s health, so, in order to be able to apply machine learning and AI, we need to tailor the AI to small data sets. That's a big part of the methodology that will be used in this project.”
Bias is by now a widely recognized problem that can affect how well an ML model works, especially for those in minority or underrepresented communities. Miolane is making the audit and elimination of such bias a centerpiece of her research, saying, “We will dedicate resources to ensure that these models are explainable, and that they accurately represent diverse populations.”
The first step in achieving that is to evaluate whether there is any problem with the model, that is, to report whether we're doing significantly worse on one sub-population or another. Miolane explains. “And if there is a problem, then we can use various proven techniques to weight the data so that the algorithm predicts correctly for the whole population.”
This grant was based on observing and predicting neuro-anatomical changes, which involves more than the anatomy of brain structures. Carmona, for instance, looked at fifteen different brain structures in the hippocampus, the amygdala, and the hypothalamus, and computed that the volume of each changed throughout and after pregnancy, decreasing and then increasing again.
“We want to bring this volumetric analysis to life by providing a model that you will be able to access in your web browser to see how the 3D structure of the brain changes. You’ll be able to say, ‘Show me how the brain looks twelve months into the future,’ or ‘Show me how it looks if the hormone levels are really high,’” Miolane says. “You’ll then see in 3D the shape of different brain structures evolving through time. For example, you can explore what the hippocampus is meant to do — how it shapes memory, learning, and navigation — and watch it evolve before your eyes throughout pregnancy. The interface becomes a window into the brain’s deeper function. We hope to provide this information in interesting detail, so that pregnant women might want to check up week by week, using it as a tool for education, empowerment, and ultimately, predictive care.”
COE News – "Leveraging AI to Study Women’s Brains During and After Pregnancy"