By Jordan Johnson
Artificial intelligence is already reshaping oncology. Bridge Oncology examines how health systems can move beyond AI hype to improve patient access, clinical workflows, workforce capacity, governance, and total cost of care.
Artificial intelligence is no longer a future concept for healthcare. It is already embedded in the systems patients and clinicians use every day.
AI is analyzing images, assisting with treatment planning, documenting visits, drafting patient communications, prioritizing work queues, translating complex information, supporting clinical decisions, and helping patients navigate appointments, test results, and medical bills.
One statement from a recent healthcare AI presentation captured the current moment perfectly:
Clinicians are adopting AI at the speed of trust. Patients are adopting AI at the speed of desperation.
That tension should command the attention of every oncology leader.
Clinicians are appropriately cautious. They understand that an AI-generated answer can appear polished and authoritative while being incomplete, outdated, biased, or simply wrong. Patients, however, may be frightened, overwhelmed, waiting for a returned call, trying to understand a pathology report at midnight, or wondering why their treatment has not been authorized.
Many patients are not waiting for healthcare organizations to complete their AI strategies. They are already using consumer AI tools to understand their diagnoses, test results, treatment options, and financial responsibilities.
The question is no longer whether AI will enter oncology. It already has. The real question is whether health systems will deploy it intentionally enough to improve care—or allow it to become another expensive, fragmented layer in an already complicated system.
AI Is Not a Single Technology
Healthcare leaders often discuss “AI” as though it were one product or capability. A more useful framework separates AI into four categories.
Perceptual AI senses and recognizes. It identifies patterns in images, signals, video, and other clinical data. Oncology applications include imaging analysis, automated segmentation, acute-event detection, breast imaging support, and treatment-planning assistance. The FDA’s list of authorized AI-enabled medical devices demonstrates how rapidly this area is growing, particularly in radiology and imaging.
Generative AI creates. It drafts notes, summaries, instructions, patient messages, policies, appeal letters, and other content. Epic’s patient-facing assistant, Emmie, shows how generative and conversational AI are moving into scheduling, billing support, chart navigation, reminders, and patient education.
Agentic AI acts. Instead of merely drafting or recommending, an agentic system can perform a multistep workflow within defined permissions. In oncology, it could identify a missing authorization, locate supporting documentation, compare the record with payer criteria, route an exception for human review, and track the case until it is resolved.
Physical AI interacts with the physical world. Robotics, smart devices, pharmacy automation, imaging platforms, and AI-enabled treatment systems connect digital intelligence with the delivery of care.
These categories matter because the governance required for a tool that simplifies patient instructions is very different from the oversight required for a system that influences diagnosis, treatment selection, medical necessity, or treatment delivery.
Why Oncology Is an Ideal—and High-Risk—Environment for AI
A cancer episode produces enormous amounts of information across pathology, molecular testing, imaging, radiation oncology, medical oncology, surgery, infusion, specialty pharmacy, navigation, prior authorization, and revenue cycle.
The problem is rarely a lack of data. The problem is converting fragmented data into timely action.
AI can help oncology teams:
- Identify missing records before a consultation or simulation
- Summarize a patient’s longitudinal cancer history
- Compare clinical documentation with payer requirements
- Detect coding and authorization mismatches before they delay care
- Improve infusion scheduling and treatment capacity
- Prioritize navigation outreach based on clinical or financial risk
- Reduce repetitive documentation and in-basket work
- Translate complex treatment instructions into understandable language
- Identify patients at risk of toxicity, treatment interruption, or becoming lost between specialties
The opportunity is substantial, but so is the risk.
AI can fabricate a payer requirement, reproduce bias embedded in historical data, rely on an outdated guideline, agree too readily with a poorly framed question, or omit the one fact that changes the entire decision.
The most dangerous failure is rarely an obviously absurd answer. It is an answer that is 95% correct, confidently presented, and wrong in the one place that matters.
Leadership Sets the Conditions for AI to Deliver
Successful AI deployment is not simply a technology project. It is a leadership responsibility built on data quality, operating discipline, workforce engagement, governance, and accountability.
Five management priorities should guide healthcare AI strategy.
Centralize and govern the data
A patient’s oncology story may be spread across the hospital EHR, radiation oncology information system, treatment-planning platform, imaging archive, pathology system, infusion platform, specialty pharmacy, payer portal, and patient-reported outcome tool.
If those systems disagree about the diagnosis, stage, treatment intent, authorization status, or next step, AI will not repair the underlying governance failure. It may simply automate the inconsistency.
Before asking whether an AI tool is intelligent, leaders should ask whether the underlying data are trustworthy, timely, accessible, and clinically coherent.
Bring leaders the problem before a vendor brings the product
Healthcare organizations should resist product-led AI strategy. When a vendor defines the problem, presents the solution, and establishes the success measures, the health system may purchase an answer before agreeing on the question.
Clinical teams, operations, IT, compliance, cybersecurity, legal, finance, and patient representatives should be involved before a tool is selected. Sometimes AI will be the best answer. Sometimes the solution will be workflow standardization, integration, staffing, education, or eliminating an unnecessary step.
The objective is not to find places to deploy AI. It is to solve meaningful problems.
Leverage what the organization already owns
Many health systems already license AI-enabled capabilities through Epic, Microsoft, Workday, ServiceNow, imaging platforms, and other enterprise systems. Leaders should evaluate those capabilities before adding another point solution, interface, subscription, and source of fragmentation.
However, leveraging the existing technology stack should not become an excuse for unquestioned vendor lock-in. Every solution must still demonstrate clinical fit, interoperability, safety, performance, and economic value.
Make responsible experimentation legitimate
Organizations cannot build an AI-capable workforce through policy documents alone. Teams need approved tools, clear boundaries, practical education, and safe opportunities to learn.
If responsible use is too difficult, employees may turn to unapproved tools and create a shadow-AI problem. If experimentation is unrestricted, the organization creates privacy, security, accuracy, and compliance risks.
AI literacy must extend beyond prompting. Users need to know when to question an output, how to verify information, what data must never be entered, which decisions require qualified human review, and how to report a failure.
Demand measurable value
Every AI initiative should begin with a defined outcome, a reliable baseline, an accountable owner, and a decision rule: measure adoption and results, then scale, change, or stop.
Adoption is not the same as value. A heavily used tool can still create downstream work, shift burden to another team, worsen inequity, or increase total cost.
Oncology leaders should measure the entire workflow. Did patients begin treatment sooner? Did authorization turnaround improve? Did avoidable denials decline? Did clinicians regain meaningful patient time? Did the technology reduce treatment interruptions, improve capacity, or lower total cost of care?
Saving five minutes in one part of a workflow is meaningless if the patient still waits two weeks for treatment—or if another employee spends ten minutes correcting the output.
The Bridge Oncology Perspective
At Bridge Oncology, we believe the most valuable AI will not attempt to replace oncology professionals. It will help scarce professionals operate at the highest level of their training.
Through our work with Tensor Black and our broader network of startups, payers, attorneys, policymakers, health systems, founders, investors, and funding organizations, Bridge Oncology evaluates AI from multiple perspectives. We examine not only what the technology can do, but how it affects patient access, clinical care, compliance, reimbursement, workforce capacity, operational performance, and total cost of care.
AI should reduce the scavenger hunt for information. It should identify friction before it becomes a treatment delay. It should standardize routine work while making exceptions more visible. It should connect radiation oncology, medical oncology, infusion, surgery, pharmacy, navigation, revenue cycle, and the patient—not create another isolated platform.
Most importantly, AI should return time to the human work of cancer care: listening, explaining, reassuring, deciding, and being present.
The organizations that lead this next era will not necessarily be those with the most AI products. They will be the organizations that know where AI belongs, where it does not, how its performance will be verified, and how every recovered minute will be converted into measurable value for patients.
If your organization is looking for a strategic partner to evaluate, implement, govern, or scale AI in oncology, now is the time to see what Bridge Oncology has created.
AI is already in oncology. Leadership must now ensure that intelligence becomes action, action becomes value, and technology makes cancer care more—not less—human.