Tutorial: Drafting Your First Patient Treatment Summary Letter

This tutorial walks you through a complete session with an AI tool. The task: draft a letter for a patient who has just been seen in your clinic following a diagnosis of early-stage breast cancer. The multidisciplinary team (MDT) has agreed on a treatment plan. The patient has asked for something in writing.

From opening the tool to having a reviewed draft ready, this takes around fifteen minutes the first time. By the third time, it takes five.

You do not need technical knowledge of any kind. You need a free account at claude.ai or chat.openai.com, and a few minutes.

Step 1: Open the tool and start a new conversation

Go to claude.ai or chat.openai.com. Sign in. You will see a text box — usually at the bottom of the screen. Click inside it. Start a fresh conversation so nothing from a previous session carries over.

Step 2: Tell the tool what you are doing before you give it the clinical content

The first thing you type is an instruction about the task — not the clinical content. This is called setting the context. Type the following:

You are helping an oncologist draft a patient letter. I will give you 
a set of clinical bullet points. Your job is to turn them into a clear, 
warm patient letter in plain English. Use no medical jargon without 
explanation. Do not include specific drug names, doses, or any language 
about expected outcomes. Maximum 350 words.

Press Enter. The tool will reply acknowledging the instruction — something like "Understood. Please share the clinical details."

Step 3: Provide your clinical bullet points

Now type or paste your anonymised clinical summary. For this tutorial, use this example:

Clinical details:
- Woman, early fifties, newly diagnosed with early-stage breast cancer
- MDT met this week and agreed on treatment plan
- Plan: surgery first, followed by adjuvant chemotherapy and radiotherapy
- Breast care nurse specialist will make contact before surgery
- Pre-operative assessment appointment to be arranged
- Patient very anxious — she has a young family and is worried about 
  managing treatment alongside work and childcare
- She has asked for everything in writing

Step 4: Read the first draft critically

The tool will produce a letter in under thirty seconds. Here is the kind of thing it might produce:

Dear [Name],

Thank you for coming to see us this week. I wanted to follow up with a written summary, as you requested.

Following your recent appointment, your case was discussed at our multidisciplinary team meeting. The team has agreed on the following plan for your care:

The first step will be surgery. After you have recovered from surgery, the plan includes further treatment — chemotherapy and radiotherapy — to give you the best possible support going forward. Your breast care nurse specialist will be in touch with you before your surgery date to offer support and answer any questions you have.

We know that managing a diagnosis like this alongside everything else in your life — work, family, all the things you are carrying right now — is genuinely hard. Please know that your team is here for you at each step.

Your next appointment will be a pre-operative assessment. We will be in touch shortly with the date.

If you have any questions before then, please do not hesitate to call us.

Yours sincerely, [Your name and contact details]

Read this as a clinician. Ask yourself:

• Is every clinical fact accurate?
• Is anything missing?
• Does any sentence make a commitment or imply an outcome I have not confirmed?
• Does the tone feel right for this patient?

In this draft, you might notice the phrase "to give you the best possible support going forward." This is vague and slightly optimistic. You may want to remove or change it.

Step 5: Ask for a targeted revision

Do not start again. Tell the tool exactly what to change. For example:

The phrase "to give you the best possible support going forward" is 
vague. Replace it with something that more clearly explains that this 
is the agreed treatment plan for her specific situation.

The tool will revise only that sentence. Read the revision. If it is better, continue. If not, adjust your instruction.

Step 6: Check the emotional tone specifically

Read the letter once more, thinking about this patient. She is in her early fifties, anxious, with a young family. Does the letter acknowledge her situation in a way that feels genuine? Or does it feel like a form letter with her name at the top?

If the acknowledgement of her anxiety feels formulaic, try:

The paragraph acknowledging that she is managing a lot feels slightly 
generic. Rewrite it to feel more personal and specific to someone 
balancing family and work alongside treatment. Keep it brief — 
two sentences maximum.

Step 7: Copy the draft into your clinical system

Once you are satisfied with the draft, copy it and paste it into your letter template. Now add the patient's name and address, your own name and contact details, the date, and any local information — the breast care nurse's name and direct contact, the pre-operative assessment booking number, and so on.

None of these details ever went into the AI tool. They come from you, in your own system.

Step 8: Read the completed letter from beginning to end

Read it aloud, or at least move your lips. A sentence that is fine on screen may be hard to follow when read aloud under stress. This patient will probably read this letter more than once. Make every sentence earn its place.

What the tool does badly

It tends toward warmth in a way that can tip into sentimentality. Phrases like "you are not alone on this journey" may feel meaningful to some patients and hollow to others. Read every sentence for authenticity, not just accuracy.

It also does not know your patient. The adjustment for her specific situation — her anxiety about her young family, her wish to keep working — came from you. The tool provided structure and words. The humanity came from your knowledge of the person.

What to try next

Once this feels comfortable, try drafting a side-effect information sheet using the same approach. The workflow is identical — set the context, provide bullet points, review the draft, refine. The task changes. The method does not.

Remember: AI is a helpful assistant, not a clinician. You make the call.