Artificial intelligence has evolved from a punchline into a serious tool transforming how oncologists work, detect cancer early, and coordinate patient care. Rather than replacing doctors, AI is now handling routine tasks that consume clinical time, freeing physicians to focus on patients while improving outcomes across multiple areas of cancer treatment.

What's Actually Changing in Cancer Clinics Right Now?

The shift happening in oncology today isn't flashy. It's practical. Specialized AI tools called large language models, or LLMs, are being built specifically for cancer care. These are AI systems trained on medical literature and clinical guidelines to help doctors make better decisions at the point of care. One example, OpenEvidence, draws from journals like the New England Journal of Medicine and the JAMA Network, plus guidelines from the National Comprehensive Cancer Network, to deliver AI-generated insights that have already supported over 100 million clinical consultations from US clinicians.

"We are at an inflection point with the arrival of AI as a tool in advancing our work in oncology. We have seen a tremendous acceleration of early adoption, and yet we have only scratched the surface," said Matthew Matasar, chief of the Division of Blood Disorders at Rutgers Cancer Institute.

Matthew Matasar, Chief of the Division of Blood Disorders, Rutgers Cancer Institute

Beyond documentation, a newer category called agentic AI is beginning to coordinate multiple AI systems to handle complex tasks. These tools can support tumor boards, match patients to clinical trials, ensure care follows established guidelines, and coordinate care across teams. The key difference from earlier AI tools is that agentic AI can use generated content and call upon external tools to complete tasks on its own.

How Are Doctors Using AI to Catch Cancer Earlier?

Early detection remains one of cancer's biggest challenges. AI is now being deployed to help radiologists spot tumors faster and more accurately. In mammography, AI is being used to reduce the time spent on double-reviewing scans, a labor-intensive process. Several clinical trials are underway in the United States to test these algorithms, particularly because breast density varies across different populations.

The potential extends beyond breast cancer. A trial called PANORAMA, published in The Lancet Oncology, demonstrated that AI-assisted CT scans can detect early-stage pancreatic cancer, a disease typically caught in later stages when treatment options are limited. Researchers are also working to establish criteria for integrating AI models into RECIST assessments, a standard method for measuring tumors, to make lesion detection more scalable and consistent.

"What I hope for is to start using those models in real practice. As of now, many of the models that we use in radiology have been localized to a couple of use cases, but now, with oncology being one of those very high-stakes diseases, we have an opportunity here," explained Arturo Loaiza-Bonilla, systemwide chief of hematology and oncology at St Luke's University Health Network.

Arturo Loaiza-Bonilla, Systemwide Chief of Hematology and Oncology, St Luke's University Health Network

How to Integrate AI Into Cancer Care Workflows

• Reduce Documentation Burden: Deploy AI tools to handle clinical charting and documentation, freeing clinicians to spend more time directly with patients rather than completing paperwork.
• Accelerate Clinical Processes: Use agentic AI to function as connective tissue across the oncology care continuum, coordinating tumor boards, trial matching, and pathway adherence to remove friction points in routine practice.
• Validate Models in Real-World Settings: Move AI algorithms from controlled research environments into actual clinical practice, ensuring they work across diverse patient populations and different healthcare systems.

Why Is AI Particularly Valuable for Heart Health in Cancer Patients?

Cancer patients taking certain medications, like tyrosine kinase inhibitors, face risks of heart complications. Foundational AI models, which are built from millions of labeled medical tests, are now being developed to detect arrhythmias and other cardiac risk factors in this population. These models can identify dangerous patterns like QT prolongation, a heart rhythm abnormality that can emerge during cancer treatment.

Because electrocardiograms, or ECGs, are simple, widely available tests, foundational models trained on ECG data could be approved by the FDA relatively quickly as risk-stratification tools. Companies are already pursuing FDA validation to democratize these models across cardio-oncology, making them available to hospitals and clinics everywhere.

What Challenges Still Need to Be Solved?

The National Cancer Institute acknowledges that AI in oncology faces real obstacles. AI models can sometimes inaccurately represent broader patient populations, potentially perpetuating medical bias if training data is incomplete or lacks diversity. The field also needs better "explainable" AI, meaning systems that can show doctors why they reached a particular conclusion, making it easier to integrate these tools into clinical workflows.

Experts emphasize that the most meaningful AI deployments in cancer care won't be announced in press releases. Instead, they'll be measured by whether they reduce documentation burden, accelerate triage, speed up access to clinical trials, and help teams execute more efficiently. That's where the real value lies for patients and clinicians alike.