Researchers have described an artificial intelligence framework designed to help tailor treatment for one of the more aggressive forms of breast cancer, according to a study published in Nature.
The work focuses on HER2-positive breast cancer, a subtype driven by a protein called HER2 that fuels tumor growth. A common strategy is "dual HER2 blockade," which uses two drugs together to shut down that protein. When this is given before surgery to shrink a tumor, it is called "neoadjuvant" therapy. The challenge for doctors is that not every patient responds the same way, and there is no easy way to know in advance who will benefit most.
According to the Nature paper, the team built what it calls a "spatially interpretable" AI framework. The "spatial" part means the system accounts for where features sit within a tumor sample, rather than treating it as a single uniform blob. The "interpretable" part signals an effort to make the AI's reasoning understandable to clinicians, instead of producing an unexplained verdict — an important consideration when treatment decisions hang on the output.
The stated goal is to tailor that neoadjuvant dual HER2 blockade to individual patients, moving toward a more personalized approach to who gets which therapy and when.
The details published center on the framework itself; the source title does not specify the size of any patient group or how the tool performed against current practice.
Why it matters: matching aggressive cancers to the right drugs sooner could spare patients from treatments unlikely to help them — and AI that can explain its reasoning is more likely to earn a place in the clinic.