What to give ChatGPT before asking about lab results

Before asking ChatGPT or another AI tool about lab results, gather the actual lab values, units, dates, reference ranges, medications, symptoms, relevant diagnoses, recent changes, and the specific question you want help organizing. Treat the answer as a draft for preparation: useful for summaries, patterns to ask about, and appointment questions, but not as a diagnosis, treatment plan, medication instruction, or urgent-care triage.

If your labs are scattered across portals or PDFs, start with the record work first. The companion guide on how to organize years of blood test results is the better first step when you do not yet have the results in one place.

Decide what job you want the AI to do

Give the AI a bounded task before you paste any results. "Explain my labs" is too broad. It invites the answer to drift into medical advice you still need to verify.

Better jobs are narrow and practical:

• Turn these results into a short appointment summary.
• List what context is missing before I talk to my clinician.
• Compare these values over time without diagnosing me.
• Draft questions I can ask at my next visit.

You can also state what you do not want: "Do not diagnose me, do not suggest medication changes, and do not tell me whether to seek urgent care. Help me organize the information and prepare questions for my clinician."

Check privacy before you paste

Lab results are sensitive health information. Before pasting them into any AI tool, remove identifiers that are not needed for the task: full name, date of birth, address, medical record number, account number, insurance identifiers, and exact facility or clinician names unless those details are essential to the question.

Do not assume every health app or AI chat is protected by HIPAA. HHS explains that HIPAA can apply differently depending on who provides the app, who receives the data, and the relationship between the app and a covered health care provider. The FTC's mobile health app tool also explains that privacy and security obligations depend on what an app collects, how it combines data, and how it shares that information.

For ChatGPT specifically, review the tool's current data controls and terms before entering personal health information. OpenAI says outputs may not always be accurate and should not be treated as the sole source of truth or a substitute for professional advice. That is another reason to keep prompts focused on organization and questions, not medical decisions.

Start with the exact lab details

Copying one isolated number into a prompt removes the context that makes the result easier to discuss. MedlinePlus explains that lab results depend on the test, reference range, and clinical context, and that normal or abnormal flags are not the whole picture. A better input is a compact list with the test name, value, unit, reference range, lab provider, and collection date.

For each result, include:

• Test name exactly as written on the report
• Value and unit
• Reference range from that report
• Collection date, not just the upload date
• Report date, if different from collection date
• Lab provider when you have results from more than one source
• Source type: original PDF, portal export, screenshot, typed copy, or clinician note
• Prior values for the same test when you are asking about change over time
• Whether the result was flagged high, low, abnormal, or not flagged

Keep units and reference ranges attached to every value. The AI needs to know which range belongs to the specific report you are discussing.

If the same marker appears across Quest, Labcorp, a hospital lab, and a specialist clinic, label the source. For that narrower problem, use how to compare Quest and Labcorp results over time before asking an AI to summarize the trend.

Preserve the timeline

If you are asking about a change over time, give the AI the sequence in date order and say what you want compared.

For example, instead of pasting only the newest result, write:

• January 12, 2025: test name, value, unit, reference range, lab
• July 8, 2025: test name, value, unit, reference range, lab
• January 20, 2026: test name, value, unit, reference range, lab

Then ask for a bounded comparison: "Please summarize what changed, what stayed similar, and what questions I should ask my clinician. Do not infer a diagnosis." For recurring biomarkers, read how to track ferritin, ApoB, and A1c over time for a workflow that keeps interpretation separate from organization.

Add the surrounding context

The goal is not to make the AI decide what is wrong. The goal is to give it enough background to help you organize what to ask next.

Include the pieces that change how a clinician would frame the conversation:

• Why you are looking at the labs now
• Symptoms or concerns you are trying to understand
• Current medications and supplements
• Known diagnoses or relevant medical history
• Recent changes such as illness, procedures, diet changes, or medication changes
• What your clinician has already told you, if anything

Then compress it into a short context block:

• Reason for looking: annual physical follow-up, fatigue discussion, medication monitoring, or another concrete reason
• Current question: what you want to understand or prepare for
• Relevant history connected to the question
• Current list: medications, supplements, and recent changes

If you are also preparing for a new clinician, the article on how to prepare a health summary for a new doctor covers the broader one-page summary. The AI prompt should be a smaller version of that same idea: enough context to organize the conversation, not enough noise to bury the question.

Mark what is missing

Do not make the prompt look more complete than it is. If one lab panel is missing, if a portal only shows summaries, or if an older result came from memory instead of the original report, label that uncertainty.

Useful phrases include:

• "I do not have the original report for this older result."
• "This value came from a portal summary, not a PDF."
• "I am missing the reference range for this result."
• "I do not know whether these two test names refer to the same exact assay."
• "Please list what information is missing before I discuss this with my clinician."

If missing files are a recurring problem, use a missing-record log so the gaps do not disappear between portal messages, phone calls, and downloads.

Use a prompt that keeps the output bounded

Use prompts that keep the output practical and bounded:

1. "Can you help me organize these results into a short summary for my clinician?"
2. "What context is missing before I discuss this with my clinician?"
3. "What are reasonable questions to ask at my next appointment?"
4. "Can you point out which values changed over time without interpreting them as a diagnosis?"

Here is a practical prompt pattern:

"I am going to paste lab results and context. Please help me organize them for a clinician visit. Do not diagnose me, recommend treatment or medication changes, or decide whether anything is urgent. Summarize the results, list missing context, identify questions worth asking, and flag unclear source information."

Then paste the results in sections:

• Goal of the conversation
• Lab results with dates, units, ranges, and sources
• Prior values if relevant
• Symptoms or concerns
• Current medications and supplements
• Known diagnoses or relevant history
• What you want to ask next

If the output gives you a confident medical conclusion, redirect it. Ask for a safer rewrite: "Please reframe this as questions and context to discuss with my clinician rather than conclusions."

You can also ask for uncertainty: "List what you can and cannot infer from this information, what context is missing, and what I should verify with my clinician." If the answer mentions general medical background, verify it against trusted sources such as MedlinePlus, NIH, a professional medical society, your original report, or your clinician's instructions. Do not assume source links supplied by an AI tool are real until you check them.

Check the output before you use it

Do not forward an AI-generated summary without checking it against the source records:

• Did it copy every value, unit, date, and reference range correctly?
• Did it mix up collection date and report date?
• Did it treat a missing range as known?
• Did it merge results from different labs without labeling the source?
• Did it imply a diagnosis, medication change, or urgency that your clinician has not confirmed?
• Did it leave out your actual concern?
• Did it cite a source that actually says what the answer claims?

You can also ask the AI to audit itself in a limited way: "Please list which parts came directly from the lab report, which came from my notes, and which are questions or uncertainties." That does not prove the answer is correct, but it can make the draft easier to inspect.

Avoid asking the AI to decide whether you have a condition, change medication, ignore a symptom, or replace clinical care. If something feels urgent, contact a qualified clinician or local emergency service instead of waiting on an AI answer.

Keep a reusable record

If you plan to ask AI tools or clinicians about labs more than once, build a reusable health context instead of rebuilding the prompt every time. Keep your lab history, notes, medication list, timeline, and open questions in one place so each conversation starts with the same organized baseline.

Libby is built for that record-keeping layer: a place to organize your health context, prepare clearer summaries, and bring better questions to the next conversation. It does not replace clinical care or interpret results as a diagnosis. It helps keep the source material and the questions in one place so the next AI conversation or clinician visit starts from a cleaner baseline.

Quick checklist before you ask ChatGPT about labs

Before you paste a prompt, check that you have:

• The exact test names from the report
• Values, units, reference ranges, and flags
• Collection dates
• Lab provider or source
• Prior values when asking about trends
• Symptoms, concerns, diagnoses, medications, and recent changes that matter to the question
• A clear statement of what you want the AI to do
• A clear statement that the output should not diagnose, treat, change medication, or triage urgent symptoms
• A plan to verify important points with a clinician

When in doubt, ask for organization first and interpretation later with a professional. A well-prepared prompt should leave you with better questions, a cleaner summary, and fewer loose ends, not a final medical answer.

References

• MedlinePlus: How to understand your lab results
• HHS: Health app use scenarios and HIPAA
• FTC: Mobile health apps interactive tool
• OpenAI: Terms of use
• OpenAI Help Center: Data controls FAQ