A new arXiv preprint introduces CoRe-WM, a conditioned response-state world model designed to predict treatment response in breast cancer using longitudinal DCE-MRI. The model targets pathologic complete response (pCR) after neoadjuvant chemotherapy, a key prognostic indicator in aggressive breast cancer.
The abstract notes that current prediction approaches often rely on pretreatment imaging alone, and the sentence is cut off in the source, but the implication is that they miss the patient's evolving response during therapy. CoRe-WM appears to address this by incorporating response-state information over time.
Because the abstract is truncated, full technical details are not yet available. However, the proposed approach signals a shift toward using serial imaging to capture dynamic changes, potentially improving prediction accuracy for patients undergoing neoadjuvant treatment.