AI moves deeper into oncology workflows and screening

AI is becoming a more practical part of oncology care as clinicians test specialized LLMs, automated documentation, trial-matching tools, and agentic systems that can coordinate tasks across workflows. The National Cancer Institute describes AI as an “unprecedented opportunity” for cancer research and care, while emphasizing the need for clinical validation, explainable systems, reproducibility standards, and safeguards against bias from incomplete or nondiverse data.

Emerging uses are most visible in imaging and screening. AI-assisted mammography may reduce double review and support clinical trials, while CT-based models are being explored for earlier detection of cancers such as pancreatic cancer. At the 2025 San Antonio Breast Cancer Symposium, a transformer-based model showed potential for predicting recurrence after breast cancer treatment, and multimodal systems combining imaging, clinical, and molecular data outperformed the Oncotype DX 21-gene recurrence score alone.

Foundational models are also being developed for cardio-oncology and pathology, including ECG-based risk tools and slide-analysis systems that could help address shortages of pathologists in low-resource regions and rural US areas. Broader deployment will depend on governance, clinician oversight, evidence alignment, and integration into daily systems so AI can improve speed, access, and execution without adding misinformation or workflow friction.