H1 title of the lesson

In this lesson, you’ll learn a simple formula for writing perfect prompts, along with real examples of how AI is already being used in medical consultations.

Let’s get started!

5 elements for the perfect prompt:

After studying expert sources, our team identified what we now call the perfect prompt formula, composed of five key components:

Task, Context, Example, Persona, and Format.

1. Task

This is the most important component to consider when writing your prompt. It defines what you expect the AI tool to return as a response. Experts recommend always starting with a verb.

👉 Example: Imagine you need to explain hypertension to your patients clearly and simply. Your task would be:

✨ Prompt: "Explain hypertension in an easy-to-understand way."

Here, we start with the verb “explain,” which immediately signals that we are asking for an explanation.

You can also combine multiple tasks in the same prompt. For example:
Research the definition of hypertension in medical dictionaries and summarize it in simple, easy-to-understand language.

2. Context

Providing context means giving background information. These can be facts, figures, or details that help the AI understand your request better.

Continuing with the hypertension example, you could add context. Your prompt might look like this:

✨ Prompt: "I am a cardiologist working with patients of all ages, and I want to explain hypertension in an easy-to-understand way. Give me some ideas on how to do it."

3. Examples and references

Just as humans learn better with examples than theory alone, this is also very helpful for AI.

Examples help LLM systems improve the quality of results because they serve as clear demonstrations for the AI.

✨ Prompt: "I am a cardiologist working with patients of all ages, and I want to explain hypertension in an easy-to-understand way. Give me some ideas based on this example [here you paste a reference you like in terms of structure, logic, or tone]."

4. Persona

Not essential for every prompt, but especially useful when addressing a specific audience.

Continuing with the cardiologist example, the intended audience could be:

✨ Prompt: "I am a cardiologist and work with patients aged 20–60 with different educational backgrounds. Explain hypertension in an easy-to-understand way."

5. Format

Also optional, but very helpful when you want the response in a specific structure.

For example, if you are sending post-care recommendations to a patient recently treated for severe abdominal pain, you can ask the AI to provide the response as a checklist, like this:

✨ Prompt: "Provide post-care recommendations for a patient recently treated for severe abdominal pain. Give the response in a clear checklist format with check marks."

These are the five key components that can significantly improve your prompts.

Extra tips to help with your prompts:

Practical AI applications in the consultation (part I)

Now that you know more about the tools and how to structure good prompts, let’s look at some real AI use cases in your daily medical practice!

Important note: Our goal is not to cover complex scenarios, like hospitals analyzing medical images and detecting early cancer signs with AI. Here you’ll find practical, accessible, and real cases to integrate AI intelligently and responsibly into your daily practice. Let’s start!

Case 1: AI for patient communication

One of the most common ways doctors use AI is to help explain things to patients, like pre-operative instructions or medication side effects. This saves time and ensures your explanations are clear and consistent.

👉 Example: Let’s see how AI can help generate a simple explanation of chronic kidney disease for a newly diagnosed patient. You can use a prompt that emphasizes tone, format, and audience, like this example.

✨ "Prompt: You are a healthcare professional. Explain chronic kidney disease to a newly diagnosed adult patient. Write a brief, empathetic, and easy-to-understand explanation covering:

-What causes the disease
-Common risk factors
-Typical symptoms
-Treatment options (including medications)
-Recommended lifestyle changes
-Other influencing factors
-Follow-up care and next steps

Provide the information in clear bullet points, highlight positive aspects of treatment, and avoid medical jargon."

Case 2: AI for preparing clinical documentation

There are also AI tools that transcribe your conversation with the patient and instantly turn it into structured clinical notes, such as Noa Notes from Doctoralia. This means less typing and standardized clinical documentation.

👉 Example: For a first visit note, I can ask Noa Notes to organize my summary with basic patient information, reason for the visit, medical history, lifestyle, and details from the clinical examination. For each, you can add subtopics on what details to highlight.

✨ "Prompt to create your first visit clinical note:
Patient information
-Full name
-Age, sex, and date of birth
-Contact details (phone, email)
-Occupation

Reason for visit
-Main symptoms
-Onset, duration, and intensity (0–10 scale)
-Aggravating or relieving factors

Medical and family history
-Chronic diseases (e.g., diabetes, hypertension, cardiovascular diseases)
-Previous surgeries
-Current and past medications
-Allergies
-Relevant family history
-Hospitalization history

Lifestyle and habits
-Alcohol, tobacco, and substance use
-Physical activity (type and frequency)
-Sleep quality
-Dietary habits

General physical examination
-Vital signs (blood pressure, heart rate, respiratory rate, temperature)
-Overall patient condition
-Abdominal inspection and palpation
-Cardiac and pulmonary auscultation
-Basic neurological and musculoskeletal assessment"

Diagnosis
-Diagnostic hypotheses
-Primary and differential diagnoses
-ICD-10 codes

Treatment plan
-Prescribed medications
-Lifestyle recommendations
-Referrals (specialists, lab/imaging tests, mental health support, etc.)
-Follow-up planning

Follow-up
-Next appointment date
-Key clinical observations

Case 3: AI for answering routine patient questions

AI tools can also help answer routine patient questions. Chatbots on WhatsApp, websites, and even phone calls are becoming common to manage and automate this initial patient triage.

👉 Example: You can use prompts to help define the script and message flow.

✨ "Prompt: Create a friendly and concise chatbot conversation flow for a medical clinic website that guides patients through the following actions:

-Book an appointment
-Cancel or reschedule an appointment
-Check service prices
-View accepted insurance plans
-Get directions to the clinic
-Prepare for the first visit

Additionally:
Include conversation scripts and message variations for each case. Ensure the tone is professional yet friendly. Include handling of edge cases and common FAQs (e.g., “I don’t know the specialty,” “What if it’s urgent?”). The chatbot should allow users to change the topic or ask additional questions at any time."

In the next class, we’ll show more use cases, the risks of AI hallucinations, and some practical tools. See you then!