Summarising Complex Case Histories Before a Review

The Problem

Some of your most demanding patients are not the most unwell — they are the ones with the longest, most tangled histories. A patient who has been seen across multiple teams over ten years, with a combination of PCOS (polycystic ovary syndrome), insulin resistance, mood disorder, and a recent incidental adrenal finding, may have hundreds of lines of clinic letters, blood results, imaging reports, and correspondence.

Before a review appointment you need to hold all of that in your head. Often you do not have the thirty minutes it would take to read everything from scratch. You skim, you hope you have not missed something, and you know that a structured one-page summary would help — if someone had time to write it.

How AI Helps

If you have a typed or copy-pasteable version of the case history — such as a letter you have written before, or a structured clinic note — an AI tool can help you produce a summary quickly.

You paste in the clinical information (anonymised, with no identifiable patient details), describe what kind of summary you need, and the AI organises it into a structured format. You then check it, add anything missing, and use it as your pre-consultation brief.

This is not the AI reading the patient record. You are the one selecting what information to include. The AI is helping you organise and present it.

A Real Example

Dr Simons is preparing to see "Ms R", a 33-year-old with a complex history. He has written two long clinic letters about her in the past two years. He copies the relevant clinical content from those letters into a plain text document, removes her name, date of birth, and any other identifying details, and replaces them with placeholders ("33-year-old woman").

He then pastes this anonymised text into an AI tool with the following instruction:

"Here is a clinical summary of a patient with multiple endocrine conditions. Please reorganise this into a structured one-page brief with the following sections: current diagnoses, relevant history, current medications (class only, no doses), outstanding investigations, and questions to address at today's review. Use short bullet points."

The AI produces a clean structured brief. Dr Simons reviews it, notices the AI has missed a comment about adrenal imaging follow-up buried in the second letter, adds that manually, and has his one-page brief ready in about four minutes instead of twenty.

Try It Yourself

Below is an anonymised clinical summary for a patient with {{main_condition}} and {{secondary_conditions}}. Please reorganise this into a structured pre-consultation brief with these sections:
1. Active diagnoses
2. Relevant history (brief)
3. Current treatments (describe by type, not specific drug names or doses)
4. Outstanding results or investigations
5. Key questions to address today

Use short bullet points. Do not add any clinical information that is not in the text I have provided.

[Paste your anonymised clinical text here]

Before pasting, remove or replace: patient name, date of birth, NHS or hospital number, address, GP name, and any other detail that could identify the patient.

Things to Watch For

The AI can only work with what you give it. If you leave out important information when copying your notes, the summary will be incomplete. It cannot read your electronic records — you are the bridge.

Do not paste anything that identifies the patient. This is the most important rule. Even clinic letter headers with initials, hospital numbers, or GP practice names should be removed before pasting into any AI tool.

The AI sometimes adds detail it has invented. This is called "hallucination." The prompt above asks it not to add information not in the text — but always check the output against the source. If something in the summary is not in what you pasted, remove it.

Summaries are not clinical decisions. The summary is a reading aid. It does not replace clinical judgment about what matters most for that patient today.

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