AI Medical Scribe for Oncology: Treatment & Follow-up Notes

Oncology notes are among the densest in medicine: staging, regimen and cycle, performance status, toxicity grading, restaging scan summaries, and the plan that follows from all of it. An ambient AI medical scribe drafts these notes from what you say during the visit, then hands the draft back about two minutes after the patient leaves. One boundary matters more here than anywhere: the scribe captures what you say about a scan read, but it never reads the image or interprets the study. The reading stays yours. You review the draft, confirm the codes as suggestions, and sign.

Key takeaways

  • The scribe drafts oncology treatment and follow-up notes about two minutes after the visit; you review and sign.
  • It captures what you say about a scan; it does not read images or interpret studies. The reading stays yours.
  • Chemo cycle, ECOG status, and toxicity grading come from your dictation, not from any inference.
  • ICD-10 suggestions are starting points only, given how site- and behavior-specific oncology coding is.
  • Audio is discarded the moment the note drafts. Nothing is recorded.
2min

From end of visit to drafted note

$79/mo

Assist tier, per clinician, annual billing

0

Images the scribe reads or interprets

Source: Patient Square product specifications, 2026.

How an ambient scribe fits an oncology visit

An oncology visit moves through a lot of ground. You review how the patient tolerated the last cycle, check toxicities, look at the latest labs and any restaging imaging, decide whether to continue, adjust, or hold treatment, and set the next cycle or scan. A follow-up in surveillance is lighter but still records the interval, the exam, and any new symptoms.

The scribe listens through the visit. When you say the patient is on cycle 3 of the regimen and tolerated it with grade 1 nausea and grade 2 fatigue, that goes in the note. When you summarize the restaging CT as a partial response with the liver lesion down from 4 to 2 centimeters, that summary lands in the note as your reading. Two minutes after the patient leaves, you have a draft to review.

Because so much of the oncology story is longitudinal and image-dependent, your narration carries the note. The scribe holds nothing between visits and sees nothing you do not say.

The scan boundary, stated plainly

This is the line that matters most in oncology. The scribe does not open the CT. It does not read the PET. It does not interpret the MRI. It has no imaging capability at all.

What it does is capture your words about the scan. When you dictate that the restaging scan shows a partial response, or stable disease, or progression with a new lesion, the scribe writes down that you said it. The interpretation is yours, spoken by you, recorded as your reading. If you say the liver lesion measured 4 centimeters on the prior scan and 2 now, both numbers are in the note because you stated them.

Why labor this point? Because a scribe that appeared to interpret imaging would be making a clinical claim it has no business making, and in oncology that claim drives treatment. The design keeps the reading firmly with the physician. The scribe is a stenographer for your assessment, not a radiologist.

The same holds for pathology and molecular results. You state the finding, the scribe records it. It does not read a path report or a genomic panel and tell you what it means.

What an oncology scribe must capture

The vocabulary is specific and the stakes are high, so accurate capture of your spoken terms matters.

Note elementWhat you say out loudWhat the scribe drafts
Staging”Clinical stage T2 N1 M0, stage IIB”Stage as you dictated it
Regimen and cycle”Cycle 3 of the current regimen, day 1”Treatment and cycle
Performance status”ECOG 1, ambulatory, restricted in strenuous activity”Performance status
Toxicity grading”Grade 2 fatigue, grade 1 neuropathy, no neutropenic fever”Toxicities with grades
Restaging summary”CT shows partial response, liver lesion 4 to 2 cm”Your scan read, as stated
Tumor board narration”Discussed at tumor board, consensus to continue current line”Multidisciplinary discussion
Plan”Continue for two more cycles, restage after cycle 5”Plan with restaging point

Toxicity grading matters for both safety and coding, and it is easy to leave vague. If you grade the neuropathy and the fatigue and say so, the note carries the grades. If you say “some fatigue,” that is what the note says, which is why stating the grade out loud is worth the two seconds.

Tumor board and restaging, in the record

Multidisciplinary discussion is part of oncology care and part of the note. When you say the case was reviewed at tumor board and the consensus was to continue the current line, that discussion goes in the note as you described it. The scribe does not attend the board or summarize other clinicians’ views. It records your account of the outcome.

Restaging visits are where the scan boundary and the longitudinal problem meet. The scribe has no prior scan on file, so the comparison you want in the note is the comparison you speak. State the prior measurement and the current one and your assessment of the response, and the note reflects all three as your reading.

ICD-10 for oncology: suggestions to start from

Oncology coding is among the most specific in ICD-10, with site, laterality, and behavior all encoded. The scribe offers suggestions from what you said, but here more than anywhere they are a starting point.

A lung primary might surface a C34 family suggestion; a secondary in the liver, a C78.7 suggestion; a personal history of cancer, a Z85 suggestion. Every one of these needs you to confirm the exact site, side, and whether the code reflects primary, secondary, or history. The scribe cannot resolve those distinctions from a spoken sentence, so it hands you a candidate and you finalize it. Our explainer on how ICD-10 suggestions are generated walks through why oncology suggestions in particular need review.

Supportive medications and the prescription draft

Oncology visits generate supportive prescriptions constantly: antiemetics, growth factors, pain management, prophylaxis. When your plan includes one, the scribe drafts it from what you said, writing out the drug and dose so you don’t retype the order.

The scribe does not screen the draft for interactions or dosing. In oncology, where patients are on complex regimens and polypharmacy is the norm, that interaction read is exactly the check you already run on every order, and it stays yours. The draft never goes to a pharmacy and there is no e-prescribing link. You review it against the regimen, adjust what needs adjusting, and act on it in your own prescribing workflow. See our post on prescription draft safety for how the draft works and why the clinician is the safety check.

Security and ownership in a high-sensitivity specialty

Oncology records are as sensitive as health data gets. Audio is processed in memory and discarded the moment the note drafts, so no recording of the visit exists afterward. Notes are encrypted in transit with TLS 1.2 or higher and at rest with AES-256, they belong to your practice, and you can export or delete them at any time.

We offer a signed BAA to every customer, our safeguards map to the HIPAA Security Rule, and our SOC 2 Type II audit is underway. If compliance gates your decision, start with the security page and our detail on BAAs and consent.

When it earns a place in oncology

Oncology notes are long and the visits are cognitively heavy. If charting is stealing time you would rather spend with patients or on treatment planning, the trade is worth weighing: narrate your assessment as you already do, and the note drafts for your review, with the scan reading always kept as yours.

Pricing starts from $79 per clinician per month on the annual Assist plan, then steps up to the Copilot and Autopilot tiers, which add a bundled AI EHR, messaging, and receptionist and follow-up automation. On the scribe-only plan it runs alongside your oncology EHR and hands you an export; it does not write back into your system. See our comparison of AI medical scribes, the cardiology scribe overview if you co-manage cardiac toxicities, and how to evaluate an AI scribe before committing to any vendor.

Want to see it on a real treatment visit? Book a demo. Pricing is on the pricing page.

FAQ

Common questions

Does the scribe interpret my patient's scans?

No. It captures what you say about the scan read. If you say the CT shows a partial response with the liver lesion down from 4 to 2 centimeters, that is what the note reflects. The scribe does not read images or interpret studies. The reading stays yours.

Can it document a chemotherapy cycle?

Yes, from your dictation. If you say the patient is on cycle 3 of a regimen and tolerated it with grade 1 nausea, that lands in the note. You confirm the regimen and toxicity grading and sign.

Will it capture staging?

It captures the staging you state, for example a TNM stage you dictate. It does not assign a stage on its own. Staging is a clinical decision you make and the scribe records.

How does it handle ICD-10 for cancer?

It offers ICD-10 suggestions based on what you said. Oncology coding is specific about site, laterality, and behavior, so treat every suggestion as a starting point you confirm, not a final code.

Does it work with my oncology EHR?

It runs alongside any system. You copy or export the note. No integration project.

Sources

  1. National Cancer Institute (fetched July 2026)
  2. American Society of Clinical Oncology (fetched July 2026)