Day 2 of 20 · AI for Healthcare

Your AI Toolkit for Healthcare

⏱ 5 min 📊 Beginner

Yesterday you saw the time-saving potential. Today you're getting set up. By the end of this lesson, you'll have three AI tools ready to go and a clear understanding of how to use them safely in a healthcare context.

The tools are free (or have generous free tiers), they require no technical background, and you can start using them immediately. But before we dive in, we need to address the elephant in the room: patient data and privacy.

Your AI toolkit: ChatGPT for drafting, Claude for reasoning, and Perplexity for research — all used with privacy-first habits.

The golden rule — never enter real patient data

Before you touch any AI tool, burn this into your memory: never paste identifiable patient information into a public AI tool. No names, no dates of birth, no MRN numbers, no addresses. Not in ChatGPT, not in Claude, not anywhere that isn't explicitly approved by your organisation's IT and compliance team.

This applies even if you think the data is "de-identified." Unless your organisation has a BAA (Business Associate Agreement) with the AI provider and has approved the workflow, treat every public AI tool as a non-HIPAA-compliant environment.

What you can do safely:

- Use fictional patient scenarios to build and test your prompts

- Use de-identified, generalised clinical details (e.g., "62-year-old male with CKD stage 3b")

- Draft templates that you later fill in with real data inside your EMR

- Ask general medical knowledge questions that contain no patient data

This course teaches you to build prompt templates and workflows using fictional examples. You then apply those templates inside your secure clinical environment.

Tool 1 — ChatGPT (your documentation workhorse)

What it is: OpenAI's conversational AI. The most widely used general-purpose AI tool.

Best for: Drafting clinical notes, patient communication, referral letters, and structured templates.

How to get it: Go to chat.openai.com and create a free account. The free tier (GPT-4o mini) is powerful enough for everything in this course. GPT-4o (paid) is better for complex clinical reasoning but isn't required.

Healthcare tip: Create a dedicated "Healthcare" folder in ChatGPT to keep your clinical prompts organised. You can save prompt templates as conversation starters.

Tool 2 — Claude (your reasoning partner)

What it is: Anthropic's AI assistant. Excellent at nuanced analysis, long documents, and careful reasoning.

Best for: Summarising lengthy patient records, analysing complex cases, and tasks where you need careful, step-by-step thinking.

How to get it: Go to claude.ai and create a free account. Claude's free tier is generous and handles long documents exceptionally well.

Tool 3 — Perplexity (your research assistant)

What it is: An AI-powered search engine that provides sourced answers with citations.

Best for: Looking up clinical guidelines, drug information, recent research, and evidence-based answers with references you can verify.

How to get it: Go to perplexity.ai and create a free account. The free tier gives you ample searches for daily clinical reference.

Knowledge Check

Which of the following is safe to enter into a public AI tool like ChatGPT?

A

A fictional patient scenario to build and test a clinical note template

B

A patient's name and date of birth along with their symptoms for a quick note draft

C

A de-identified case that still includes the patient's MRN number

D

A patient's full history copied from the EMR with only the name removed

The only safe approach with public AI tools is to use completely fictional scenarios or fully de-identified, generalised clinical information. Even partial identifiers like MRN numbers or specific date combinations can compromise patient privacy. Build your templates with fake data, then apply them in your secure EMR.

Setting up your workflow — the template method

Here's the approach that makes AI both fast and safe in clinical practice:

Step 1: Build the template in AI — Using a fictional patient, craft your prompt until the output format is exactly what you need. Save this prompt as a template.

Step 2: Apply the template in your EMR — When you have a real patient, fill in the template variables inside your secure clinical system. Many EMRs now support AI-assisted note generation natively.

Step 3: Review and sign — Always read the output. Check for accuracy, hallucinations, and tone. Then sign it as your own work.

This three-step method gives you the speed of AI with the safety of never exposing patient data. It's the workflow we'll use throughout this entire course.

🤖

ChatGPT

online

I want to create a reusable prompt template for writing SOAP notes. The template should have placeholders I can fill in for each patient. Can you help me design one?

Here's a reusable SOAP note prompt template with placeholders: --- **Prompt Template — SOAP Note Generator** Write a structured SOAP note for the following encounter: - **Patient:** [Age]-year-old [sex] with history of [conditions] - **Chief complaint:** [reason for visit] - **Subjective:** [patient-reported symptoms, duration, severity] - **Vitals:** [BP, HR, Temp, SpO2, RR as applicable] - **Exam findings:** [relevant physical exam findings] - **Labs/imaging:** [any results available] - **Assessment:** [your clinical impression] - **Plan:** [intended treatment, referrals, follow-up] Format as a professional SOAP note. Use concise medical language. Include ICD-10 codes if applicable. --- Save this template somewhere accessible. When you see a patient, fill in the bracketed fields with de-identified or generalised details and paste it in. Want me to create similar templates for referral letters or discharge summaries?

↻ Replay conversation

Knowledge Check

What is the purpose of the "template method" for using AI in clinical practice?

A

To create a permanent connection between ChatGPT and your clinical workflow

B

To train the AI model on your specific clinical documentation style

C

To allow AI tools to directly access EMR data for faster documentation

D

To build prompts with fictional data in AI tools, then apply them securely in your EMR

The template method keeps patient data safe. You build and refine prompt templates using fictional scenarios in public AI tools, then fill in real clinical details inside your secure EMR environment. This gives you the productivity of AI without any privacy risk.

Your setup checklist

Before tomorrow's lesson, make sure you've completed these steps:

1. Create a ChatGPT account at chat.openai.com — this will be your primary documentation tool

2. Create a Claude account at claude.ai — your go-to for complex reasoning and long documents

3. Create a Perplexity account at perplexity.ai — your sourced research assistant

4. Create a "Healthcare Templates" document — a simple notes file (Google Doc, Notion, or plain text) where you'll save your prompt templates throughout this course

5. Remind yourself of the golden rule — no real patient data in public AI tools, ever

That's it. Three accounts and a document. Total setup time: about 10 minutes. Tomorrow we'll put these tools to work with your first AI-drafted clinical note.

Knowledge Check

Which AI tool is best suited for looking up clinical guidelines with verifiable source citations?

A

Claude — because it specialises in medical knowledge and clinical reasoning

B

ChatGPT — because it has the largest training dataset of medical literature

C

Perplexity — because it provides sourced answers with citations you can verify

D

All three tools provide identical citation capabilities

Perplexity is specifically designed as an AI-powered search engine that provides answers with source citations. This makes it ideal for clinical reference queries where you need to verify the evidence. ChatGPT and Claude are better for drafting and reasoning tasks.

🧰

Day 2 Complete

"Three free tools, one golden rule: never enter real patient data into public AI. Build templates with fictional cases, apply them in your secure EMR. Tomorrow you'll draft your first AI-assisted clinical note."

Tomorrow — Day 3

Your First AI Clinical Note

Tomorrow you'll draft your first AI-assisted SOAP note using a proven prompt template — and see how fast clinical documentation can really be.

Continue to Day 3 →