An oncology multidisciplinary team needs to know which evidence supports a proposed decision, which source version was used and which question remains unresolved. An AI-generated summary is useful only when it helps the team inspect that reasoning.
This guide was refreshed on 10 November 2026. It focuses on evidence preparation and review, rather than treating every oncology information product as the same category of AI. The related oncology guideline-tool comparison provides a more detailed shortlist for guideline lookup.
Separate the evidence tasks
| Task | Resource category | What the team should be able to inspect |
|---|---|---|
| Find a society recommendation | ASCO or ESMO guideline resources | Exact document, version and relevant section |
| Establish relevant UK guidance | NICE and applicable local documents | Remit, recommendation and service context |
| Explore variant implications | A specialist knowledge base such as OncoKB | Variant, tumour context and evidence annotation |
| Understand treatment-protocol detail | The protocol approved for the service | Current local version and ownership |
| Orient to an unfamiliar question | Clinical reference or a cited AI answer | Supporting source and unresolved assumptions |
The table distinguishes information roles. It does not establish that a product has access to the patient record, integrates with your meeting system or is approved for a particular clinical use.
Guidelines: preserve the conditions around the recommendation
The ASCO Guidelines Assistant provides members with conversational access to ASCO's guideline collection. ESMO supplies society guideline resources. NICE provides the relevant UK guidance directory.
When preparing a case, retain the recommendation's setting and conditions. A tumour label without stage, biomarker or treatment-history context can make an apparently relevant reference difficult to apply. If two documents seem inconsistent, compare their scope and dates before describing a disagreement.
Keep a newly published trial distinct from a guideline update. The paper may be important to discuss while the guideline remains unchanged. A useful evidence note states that difference plainly and links both documents where relevant.
Variant knowledge bases answer a more specialised question
OncoKB's own explanation describes a precision-oncology knowledge base for the biological and clinical implications of cancer variants. That role differs from a general conversational reference or a guideline directory.
For a hypothetical molecular discussion, record the exact alteration and tumour context used to find the annotation. Then identify the question the annotation answers and what further specialist interpretation is required. A statement about a variant should not silently become a complete treatment recommendation for the whole case.
The same principle applies to any specialist database: preserve its evidence structure rather than flattening it into a confident sentence detached from the original entry.
Meeting workflow needs named ownership
NHS England's MDT streamlining guidance and the NDRS explanation of MDT recording provide relevant organisational context. A new evidence tool should fit the team's agreed process for preparation, review and recording.
Before using software to prepare evidence notes, agree who checks them, who resolves disputed sources and where the final decision is recorded. A useful demonstration should show what happens when information is missing or a source is outdated, alongside a successful retrieval.
This is also the point to establish the tool's intended use, data requirements and approved organisational arrangements. A reference licence or a vendor announcement alone does not demonstrate that patient-record integration is available to your team.
A worked preparation example
Consider a fictional trainee preparing an evidence note for an MDT teaching case. They find a trial, an international guideline and a relevant UK document. Instead of asking a model for a single answer, they create a short note with three entries: what the trial studied, what the guideline currently recommends and which local question remains open.
The supervisor can now review each claim against its source. If a required biomarker or treatment detail is missing from the fictional case, the note records that gap. The result is a more reviewable discussion than a long summary that appears to settle the question without exposing its assumptions.
Follow the unresolved question into learning
Ask iatroX can support UK source-linked reference questions. The relevant exam library and Tutor can help a learner work through a difficult distinction afterwards. CPD provides a route to a personalised account of that learning.
Keep the professional-learning record distinct from the team's clinical decision record. Describe the question you explored, how your understanding changed and what you will revisit. That makes the learning useful without presenting an educational activity as a substitute for the MDT process.
Frequently asked questions
Is a variant database the same as an oncology guideline?
They address different information needs. Preserve the database's annotation context and consult the relevant clinical guidance separately.
Can a new trial be described as an updated recommendation?
Only establish that by checking the current guideline. Keep trial findings and published recommendations distinct.
What makes an evidence note easy to review?
Link the source, record its version and relevant passage, and identify the unresolved question and responsible reviewer.