Researchers at Harvard-affiliated Mass General Brigham have developed BrainIAC, a new AI foundation model that can identify multiple neurological health risks from routine brain MRI scans using far less data than traditional systems. The model can estimate brain age, predict dementia risk, detect brain tumor mutations, and forecast survival in brain cancer patients. Trained on nearly 49,000 MRI scans, BrainIAC uses self-supervised learning to extract meaningful patterns from unlabeled data, making it adaptable across hospitals and imaging types. It outperformed several task-specific AI models, especially in complex cases or when training data was limited. Published in Nature Neuroscience, the research shows how AI can support faster diagnosis, better research insights, and more personalized care.

Insight: As AI becomes more flexible and data-efficient, students entering STEM fields will help shape healthcare innovations that directly improve lives.