AI Foundation Model Predicts Diseases From Brain Scans

BrainIAC AI model can forecast brain age, dementia, stroke, and cancer from MRI scans

Published on Feb. 5, 2026

Got story updates? Submit your updates here. ›

Researchers have unveiled an AI foundation model called BrainIAC that can predict brain age, dementia, time-to-stroke, and brain cancer from magnetic resonance imaging (MRI) scans. BrainIAC was pre-trained on over 48,900 brain MRI scans across 10 neurological conditions and outperformed traditional supervised AI models, especially in situations with limited data or high task difficulty.

Why it matters

This study showcases the potential of flexible, general-purpose AI foundation models to revolutionize medical diagnostics. Unlike narrow, single-purpose AI models, BrainIAC demonstrates the ability to tackle a wide range of brain health prediction tasks, which could lead to earlier detection and better treatment of neurological diseases.

The details

The BrainIAC model was developed by researchers at the Dana-Farber Cancer Institute, Brigham and Women's Hospital, and Harvard Medical School. It was pre-trained on 48,900 brain MRI scans covering 10 neurological conditions, including Alzheimer's, dementia, stroke, Parkinson's, and various brain cancers. BrainIAC was then able to perform a variety of tasks, such as predicting brain age, dementia, time-to-stroke, and brain cancer survival, as well as identifying genetic mutations. The researchers found that BrainIAC outperformed traditional supervised AI models, especially in situations with limited data or high task difficulty.

  • The study was published on February 6, 2026 in the journal Nature Neuroscience.

The players

BrainIAC

An AI foundation model developed by researchers that can predict brain age, dementia, time-to-stroke, and brain cancer from magnetic resonance imaging (MRI) scans.

Benjamin Kann

The corresponding author of the study, affiliated with the Dana-Farber Cancer Institute, Brigham and Women's Hospital, and Harvard Medical School.

Got photos? Submit your photos here. ›

What they’re saying

“We find that BrainIAC consistently outperforms traditional supervised models and transfer learning from more general biomedical imaging models across a wide range of downstream applications on healthy and disease-containing scans with minimal fine-tuning.”

— Benjamin Kann, Corresponding author (Nature Neuroscience)

What’s next

The researchers plan to explore incorporating additional data sources, such as omics and clinical data, to further enhance the performance of the BrainIAC model.

The takeaway

This study demonstrates the power of flexible, general-purpose AI foundation models to revolutionize medical diagnostics. By leveraging large, diverse datasets and transfer learning, BrainIAC was able to outperform traditional AI models in predicting a wide range of neurological conditions from brain MRI scans, paving the way for earlier detection and better treatment of these diseases.