ChatGPT Health Just Turned Your Scattered Medical Data Into One Brief

What you'll learn in this article:

• What ChatGPT Health is, what it connects to, and how its privacy wall actually works
• How to link Apple Health, MyFitnessPal, and medical records in under 10 minutes
• Five exact prompts for lab interpretation, appointment prep, and insurance questions
• How to read a lab panel without spiraling into a symptom-search panic
• Where this fits against Claude and Gemini health features, and what it does not do

Your health data is everywhere and nowhere. Your Apple Watch has a year of sleep and heart-rate data. Your last physical produced a PDF you saved and never opened. MyFitnessPal knows your macros. Your insurer's portal holds claims you cannot decode. None of it talks to each other, and the one time it matters, the morning of a doctor's appointment, you are scrolling through screenshots trying to remember what your cholesterol was last time.

On January 7, OpenAI shipped a fix. ChatGPT Health is a dedicated space inside ChatGPT that connects those sources and answers questions about your own data, without using those conversations to train its models. OpenAI says roughly 230 million people already ask ChatGPT about health every week. Now those questions can reference your actual numbers instead of generic web advice.

The stakes are practical, not abstract. A 15-minute doctor visit goes better when you arrive with a one-page brief of what changed and three sharp questions. A confusing lab result is less frightening when something explains each value in plain language. This guide walks through connecting your data and running five specific workflows, with the exact prompts, so you leave with a repeatable system rather than a one-time novelty.

What ChatGPT Health actually is

ChatGPT Health is a separate tab inside the ChatGPT app, not a different app and not a different subscription. OpenAI built it as a walled space: conversations you have inside Health are not used to train OpenAI's foundation models, and the space carries enhanced privacy controls compared to a normal chat. It runs on the GPT-5.2 family, the same models powering the rest of ChatGPT.

The point of the separate space is data gravity. Instead of pasting a screenshot into a regular chat, you connect a source once and then ask questions against it repeatedly. The connected sources at launch include Apple Health (on iOS, covering movement, sleep, and activity), MyFitnessPal (nutrition, macros, recipes), Function, and electronic medical records and patient portals (lab results, visit summaries, clinical history).

A note on rollout: OpenAI launched Health to a small group of early users first and is expanding access in the weeks after launch. If you do not see the Health tab yet, it is on the way, not missing.

Step 1: Open the Health space and connect Apple Health

On iPhone, open ChatGPT and look for the Health tab. Tap into it, then open the connectors panel. Connect Apple Health first, since it is the richest passive data source you already own. You will approve which categories ChatGPT can read: movement, sleep, heart rate, and activity are the high-value ones.

This single connection is what separates Health from a normal chat. Once Apple Health is linked, ChatGPT can reference your actual sleep and heart-rate trends instead of asking you to describe them.

Step 2: Connect MyFitnessPal and your medical records

In the same connectors panel, add MyFitnessPal if you track food, and connect your medical records or patient portal if your provider supports it. Medical records bring in lab results, visit summaries, and clinical history. If your portal is not directly connectable, you can still upload a lab PDF or visit summary directly into the Health space.

Connect only what you will use. If you never log food, skip MyFitnessPal. The value is in the sources you actually feed.

Step 3: Run the lab-interpretation workflow

This is the workflow most people will use first. Upload a recent lab panel (or let it pull from connected records) and run this prompt:

LAB INTERPRETATION PROMPT

Here is my recent lab panel. In plain language, tell me what each out-of-range value means, which ones matter most for someone my age and sex, and write 3 specific questions I should ask my doctor at my next appointment. Do not give a diagnosis. Flag anything that warrants prompt attention.

ChatGPT will return a value-by-value read, a short list of what matters most, and three questions. The "do not give a diagnosis" instruction keeps it in interpretation mode rather than playing doctor, which is exactly the line you want.

Step 4: Build a doctor-appointment brief

The week before an appointment, run this against your connected data:

APPOINTMENT PREP PROMPT

Using my connected Apple Health data and recent labs, write a one-page brief for my upcoming doctor visit. Include: what has changed since my last visit, the 3 trends worth raising, and 5 questions to ask. Keep it to one page and write it so I can read it aloud in the room.

You walk in with a printed page instead of a vague memory. Doctors notice, and the visit runs tighter.

Step 5: Decode an insurance or coverage question

Health data is not only clinical. Connected records can include claims and coverage patterns. Use this when a bill or coverage decision confuses you:

INSURANCE QUESTION PROMPT

Based on my healthcare usage patterns, explain in plain language what this coverage decision means, what my likely out-of-pocket exposure is, and the 2 most useful questions to ask my insurer. Do not guess at numbers you cannot see in the data.

The last clause matters. It stops the model from inventing dollar figures, a failure mode that turns a helpful brief into a misleading one.

How this compares to Claude and Gemini

OpenAI is not alone here. Anthropic followed within days with healthcare-focused features for Claude, including connectors to medical reference databases. Gemini handles uploaded health screenshots well too. The difference ChatGPT Health makes is the persistent connected space: you link a source once and keep asking, rather than re-uploading every time. For an executive who wants a repeatable system, that persistence is the practical edge.

What it does not do

ChatGPT Health interprets and organizes. It does not diagnose, prescribe, or replace a clinician, and you should not ask it to. Treat it as the analyst that preps you for the human expert, not the expert. Sensitive data is involved, so use it on your own account and review what you connect before you connect it.

Action Steps Summary

• Open the Health space: Tap the Health tab in ChatGPT on iOS; if it is not there yet, access is rolling out.
• Connect your richest source first: Link Apple Health for sleep, heart rate, and activity, then add MyFitnessPal and medical records as needed.
• Interpret a lab panel: Run the lab-interpretation prompt to get a plain-language read plus three doctor questions.
• Build an appointment brief: Generate a one-page, read-aloud brief before your next visit.
• Decode coverage: Use the insurance prompt to translate a confusing coverage decision, with numbers grounded in your data.