What will define the next 25 years of cancer research? A new Nature Reviews Cancer Viewpoint argues that oncology must move beyond describing established tumours towards predicting their future behaviour, intercepting disease before invasion and eliminating the capacity of residual cancer cells to restart tumour growth. Two concepts particularly stand out. Minimal residual disease may not be defined adequately by the number of malignant cells that remain: a small population with high “restart potential” could be more clinically important than a larger population of irreversibly damaged cells. Similarly, larger and more complex AI models are not automatically better biological models. Progress will depend on well-annotated datasets, rigorous benchmarks and cancer-specific computational tools. This is a forward-looking expert Viewpoint rather than evidence supporting immediate clinical practice change. Nevertheless, its central message is compelling: the future of oncology will depend not only on detecting and describing cancer, but on predicting how it will evolve, resist treatment and restart. ☑️nature.com/articles/s41568-0… #CancerResearch #PrecisionOncology #CancerGenomics #ArtificialIntelligence #TranslationalResearch @oncodaily @Nature @NatureRevCancer @dougflora2