Precision oncology has given physicians more ways to treat cancer. It has also created a harder problem: keeping up with the evidence needed to decide which treatment makes sense for which patient, and when that evidence changes.
At the 2026 Annual Meeting of the American Society of Clinical Oncology, more than 7,000 studies were presented across areas including pancreatic, lung, prostate and other cancers, with findings covering survival, recurrence, disease progression and quality of life. The volume is a sign of scientific progress. It is also a reminder that the traditional way of managing evidence is becoming increasingly difficult to sustain.
The problem is not that oncologists lack access to information. They have almost too much of it.
A physician trying to answer a specific question may need to move between clinical guidelines, published studies, regulatory decisions, biomarker information, clinical trials and conference findings. Each source serves a purpose, but the information is rarely organized around the exact patient sitting in front of the physician.
That creates a dangerous paradox. Medicine is becoming more precise while the evidence infrastructure supporting those decisions remains fragmented.
A diagnosis such as lung cancer can now involve numerous biologically and clinically distinct patient populations. Biomarkers, disease stage, previous treatment and molecular characteristics can all influence which therapies are relevant. The more precisely we define patients, the more precisely we need to organize the evidence that informs their care.
Clinical guidelines remain essential. Systematic literature reviews remain essential. Neither should be discarded because AI has arrived. But neither was designed for an environment in which evidence can change continuously.
...
Living Evidence
The industry should therefore stop thinking about AI in oncology primarily as a faster way to search for information. The more important opportunity is to build systems that continuously maintain the evidence itself.
This is the idea behind "living evidence": rather than treating an evidence review as a document that is completed and eventually becomes outdated, the underlying evidence base can be continuously monitored, updated and evaluated using systematic methods, as new research, regulatory decisions and clinical developments emerge.
The distinction is important because speed without rigor is not progress in medicine.
AI can scan enormous quantities of information quickly. It can identify publications, extract relevant findings and help connect information across sources. But oncology cannot rely on systems simply because they are fast. When an evidence base influences a treatment decision, the ability to trace where that information came from, how it was evaluated and whether it remains current matters just as much.
That is why the future of AI in oncology should not be about replacing the evidence hierarchy. It should be about making that hierarchy more usable.
Beyond the Clinic
The need extends beyond the clinic.
Pharmaceutical companies operate within the same expanding evidence environment. Development decisions, regulatory strategies, health technology assessments and reimbursement discussions all depend on understanding what the scientific literature says about a disease, treatment, comparator and patient population.
Those questions are often addressed through systematic literature reviews commissioned for a specific purpose. They can be rigorous, but they can also be time-intensive, particularly when the underlying literature is large and continually changing.
The opportunity is to build the evidence foundation before a new question arises.
Instead of starting from scratch every time a pharmaceutical team needs to understand a particular evidence landscape, an organization could maintain a continuously updated evidence base and then use that foundation to answer specific questions about populations, interventions, comparators and outcomes.
The systematic review does not disappear. The infrastructure supporting it becomes continuously maintained and reusable.
That shift is particularly relevant as the life sciences industry itself tries to move beyond AI experimentation and toward practical deployment. Deloitte’s 2026 Life Sciences Outlook found that 48 percent of surveyed life sciences executives identified accelerated digital transformation as an influential trend, while only 22 percent said their organizations had successfully scaled AI.
The lesson is not that life sciences needs more AI. It is that the industry needs better applications of AI to problems where the technology can produce durable value.
Evidence management is one of those problems.
Black Book Research’s 2026 State of Digital Healthcare in Oncology report, based on approximately 1,280 U.S. oncology stakeholders, identified issues including referral-to-treatment delays, fragmented pathology and genomics information, prior authorization workloads, clinical trial matching and workforce pressures among major concerns facing oncology organizations.
These challenges may appear different on the surface, but many ultimately depend on the same thing: getting the right information to the right person at the right time.
The Human Problem
This is where I believe the concept of living evidence deserves more attention.
My interest in this problem began long before the current AI boom. After more than 25 years working across pharmacy, health economics, pharmaceuticals and evidence strategy, I had seen how difficult it could be to turn a vast body of scientific research into an answer to a specific question.
While helping someone navigate a cancer diagnosis, I remember spending hours working through a lengthy clinical guideline to find information relevant to one particular concern.
The information existed. The challenge was finding the relevant evidence, understanding its context and determining whether it was still current. That experience highlighted a wider problem in healthcare. Generating scientific evidence is only part of the challenge. The evidence must also be organized, maintained and accessible enough to support the questions clinicians face in practice.
AI could help address this challenge by making it easier to identify and organize research, but speed alone is not enough. In oncology, evidence must remain traceable, scientifically rigorous and subject to appropriate human oversight.
The question is not simply how to find answers faster. It is how to ensure that the evidence informing clinical decisions keeps pace with scientific progress. That matters because oncology is not a field where plausible-sounding information is good enough.
A system has to show where its information came from. It has to account for new evidence. It has to distinguish between research findings, regulatory approvals and clinical guidance. And it has to make it possible for the user to understand the basis for the information being presented.
AI can accelerate that work, but human oversight remains critical.
Building for the Future
The same principle applies beyond clinical care. Pharmaceutical research, regulatory decisions and health technology assessments all depend on evidence that is comprehensive, current and relevant to the question being asked. Maintaining that evidence over time could reduce the need to repeatedly reconstruct the same foundations as new questions arise.
The broader challenge is to build evidence systems that evolve alongside the science, making rigorous information easier to maintain, evaluate and apply across healthcare. Healthcare has spent decades building increasingly sophisticated systems for generating evidence. We now need equally sophisticated systems for maintaining it.
The alternative is an information environment where scientific progress continually creates more work for the people responsible for interpreting it.
More publications should not automatically mean more hours spent searching. More biomarkers should not automatically mean more fragmented information. More clinical trials should not automatically mean more disconnected evidence. And more AI should not simply mean more summaries. The real opportunity is to create evidence infrastructure that becomes more current as the science changes, while remaining transparent and scientifically rigorous.
For patients, the stakes are obvious. A physician's treatment discussion depends on the quality and relevance of the information available at that moment. For pharmaceutical companies, the consequences extend into development, regulatory strategy, market access and reimbursement.
Precision oncology has made cancer treatment more individualized. Our evidence systems need to become equally specific to the patient and the decision.
The future of oncology will not be defined by who can generate the most information. It will be defined by who can turn an ever-growing body of information into evidence that clinicians and decision-makers can trust.
AI has a role in making that possible.
But the goal should never be AI for its own sake. The goal should be living evidence: continuously updated, human-vetted and structured around the decisions that matter.
If precision oncology is going to deliver on its promise, the industry needs to build an evidence infrastructure that can keep pace with the science. The data is already growing. Now the systems that make sense of it need to catch up.
Anna Forsythe is founder and CEO of Oncoscope-AI, which develops the oncology evidence technology advocated in this article.
All opinions are the author’s own.