Tutorial: Using AI to Prepare a Teaching Session on Diabetes for Junior Doctors

This tutorial shows you exactly how to use an AI tool to prepare a teaching session. We will use a realistic scenario: you have agreed to run a forty-five-minute session for foundation year doctors on recognising and managing hypoglycaemia (low blood sugar) in hospital inpatients.

You know the content. The challenge is finding time to structure it properly. This walkthrough will get you from "I need to prepare this" to "I have a reviewed outline and talking points" in about forty-five minutes, most of which is your own thinking time.

---

The Scenario

You are running a teaching session next Tuesday. The audience is foundation year 1 and year 2 doctors — enthusiastic, busy, and not yet confident with insulin management on the wards. You have forty-five minutes, no more.

You want to cover: recognition, immediate management, why it happens in hospital (insulin timing, missed meals, renal function changes), common errors, and a case to anchor the learning. You also want it to be interactive, not a lecture.

---

Step 1: Start With Your Own Thinking — Before the AI

Open a blank document and spend ten minutes writing down everything you want the session to cover. Do not worry about structure or order. Just brain-dump. Include common errors you see on the wards, the misconceptions foundation doctors have, the one thing you wish someone had taught you early.

Your raw notes might look like:

• Recognition: what counts as hypo? Different thresholds? Symptomatic vs asymptomatic
• Why hospital inpatients are at higher risk — renal function, meal timing, insulin charting errors
• Common error: treating a hypo without stopping the insulin infusion
• Common error: not rechecking glucose after treatment
• Common error: overcorrecting and causing hyperglycaemia
• Foundation doctors don't feel confident challenging insulin prescriptions
• Case: elderly woman with type 2 on insulin, post-op, missed lunch
• Interactive: ask the group — what would you do at each step?

This is the content the AI will help you structure.

---

Step 2: Write Your First AI Prompt

Open your AI tool (claude.ai or chat.openai.com). Type the following, replacing the items in brackets with your own content:

"I'm preparing a 45-minute teaching session for foundation year doctors on recognising and managing hypoglycaemia in hospital inpatients. Here are the key points I want to cover: [paste your bullet points here].

Please produce a structured session outline with:

• 3–4 clear learning objectives
• A logical section structure (with timings totalling 45 minutes)
• Key teaching points for each section (bullet form)
• One brief fictional patient case to use for interactive discussion
• Two or three common errors to highlight

Clinical focus, practical. No specific drug doses or treatment protocols. The session should feel interactive, not lecture-heavy."

---

Step 3: Review the Outline the AI Produces

The AI will produce something like this:

---

Session: Hypoglycaemia in Hospital Inpatients — Foundation Teaching Duration: 45 minutes

Learning Objectives By the end of this session, participants should be able to:

1. Recognise the clinical features of hypoglycaemia in an inpatient
2. Describe the immediate steps to take when hypoglycaemia is identified
3. Identify three common risk factors for hospital-acquired hypoglycaemia
4. Name two common management errors and explain how to avoid them

Section 1: What Is Hypoglycaemia and Why Does It Happen in Hospital? (10 min)

• Defining the threshold (note: emphasise that symptomatic patients may need treatment regardless of exact number)
• Why inpatients are at higher risk than outpatients: changes in renal function, missed or delayed meals, insulin charting errors, post-operative fasting
• Group question to open: "Has anyone seen a hypo on the ward? What happened?"

Section 2: Recognition — What You Will Actually See (10 min)

• Typical features: sweating, tremor, confusion, pallor
• The patient who doesn't look classic: elderly patients, those on beta-blockers, post-surgical patients
• Why under-recognition happens — and how it leads to avoidable harm

Section 3: The Case — Interactive Discussion (15 min)

• Fictional case: "Mrs P, 74, type 2 diabetes on insulin, post-op day 1, misses lunch. Nurse calls you because she is confused."
• Walk the group through: initial assessment, immediate action, what to do about the insulin, when to recheck, who to escalate to
• Facilitated discussion: what would you do at each step?

Section 4: Common Errors and How to Avoid Them (7 min)

• Error 1: Treating the hypo without addressing the ongoing insulin
• Error 2: Not rechecking the glucose after treatment
• Error 3: Overcorrecting and then managing rebound hyperglycaemia as a new problem

Section 5: Wrap-Up and Questions (3 min)

• Summary: three things to remember
• Open Q&A

---

This is a strong draft. Now you work on it.

---

Step 4: Edit the Outline to Match Your Thinking

Read the outline carefully. A few things to consider:

Does the timing feel right? Fifteen minutes for a case discussion might be tight if the group is engaged. You might adjust to twelve minutes for recognition and eighteen for the case.

Are the learning objectives exactly what you want? You might add "Explain why escalating concerns about insulin prescriptions is appropriate" if foundation doctor confidence is a specific goal.

Is anything missing from your original notes? You noted that foundation doctors often do not feel confident challenging prescriptions. That is not explicitly covered. Add it to Section 4 as a fourth point.

---

Step 5: Ask for Talking Points for Each Section

Now you have an outline you are happy with, ask the AI to produce speaking notes:

"For the outline above, write two to three sentences of talking-point notes for each section — what I might actually say to open each part. Conversational, not formal. No specific drug doses."

This gives you a set of speaker notes that sound like you are talking, not reading.

---

Step 6: Ask for Quiz Questions (Optional but Useful)

If you want to make the session more interactive, ask the AI to produce two or three quick quiz questions for the group:

"Write three short quiz questions (multiple choice, two options each) to use interactively during the session. Base them on the teaching points in the outline above. No drug doses."

Use these as "hands up" or verbal response questions to check understanding and increase engagement.

---

Step 7: Check Clinical Accuracy Carefully

Teaching content requires extra scrutiny. Anything you say in a teaching session will be remembered by learners and potentially applied on the wards. Read every clinical statement in the draft carefully against your own knowledge. If anything feels slightly off — for example, a statement about thresholds for treatment, or about which patients are at highest risk — adjust it yourself. The AI produces plausible language, not guaranteed accuracy.

---

Step 8: Build Your Slides

Take the section headings and key bullet points from the outline and build your slides in PowerPoint or Keynote. Each section heading becomes a slide title. The key points become slide content. The case becomes a discussion slide. Your talking points go in the speaker notes.

The AI has done the thinking structure. You do the visual design.

---

Step 9: Rehearse Against the Timing

Run through your session mentally (or aloud) against the clock. Foundation doctors ask a lot of questions — build in extra time for the case discussion if you can.

---

What You Have Just Done

You turned a brain-dump of raw clinical knowledge into a structured, reviewed teaching session in about forty-five minutes. The AI gave you structure and draft language. You gave the session clinical accuracy, local relevance, and your own voice.

The same process works for any teaching format: grand rounds, departmental updates, undergraduate teaching, and registrar training.

---

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