Turn three Apple Health trends into better checkup questions

Jordan opens the Health app before an annual visit and sees months of sleep, activity, heart-rate, and workout data. The volume is impressive and not especially useful in its raw form. In this illustrative scenario, the goal is not to make Gemini interpret a complete health record. It is to arrive with three clear observations and three better questions.

Jordan reviews the 90-day charts, writes down one trend in resting heart rate, one change in sleep duration, and one activity pattern. Dates are rounded to weeks, names and locations are removed, and the note includes only information relevant to the upcoming conversation.

Gemini helps organize the note into a one-page brief and suggests neutral questions, such as which changes are worth monitoring and what additional context the clinician needs. It does not diagnose the trends, set a target, or recommend changing medication. The AI helps prepare the conversation. The clinician handles the medical judgment.

Do not upload the raw export for this workflow

Apple lets an iPhone user export all Health and Fitness data in XML format. That export can be useful as a personal backup or for a service a person has deliberately approved, but it may contain far more information than a checkup-prep prompt requires.

Google's Gemini privacy notice says data and files provided to Gemini are processed under the account's settings and applicable terms. Google's file-upload documentation lists supported documents, spreadsheets, and other formats, but it does not establish a simple, official workflow for interpreting an Apple Health XML archive as a clinical record.

For this guide, keep the raw export off the prompt. Select and transcribe three relevant trends manually. Data minimization is the point: share the smallest amount needed to organize the appointment questions.

If a clinician or health system provides an approved upload route, use that route and follow its instructions. A general-purpose chatbot should not become the default destination for an entire health history.

Choose three trends, not three diagnoses

Open the iPhone Health app and review the last 90 days. Pick up to three categories relevant to the visit. Examples include:

• Average sleep duration by week
• Resting heart rate trend
• Daily step or activity trend
• Exercise minutes by week
• A symptom or state-of-mind log already discussed with a clinician

Write observations, not conclusions. "Average sleep duration fell from about 7 hours to 6 hours and 20 minutes over eight weeks" is an observation. "The change means a sleep disorder" is a diagnosis and should not come from this exercise.

Include context that could affect interpretation: travel, illness, a device change, shift work, a new training routine, or missing tracking days. Do not assume the wearable captured every night or that the measurement is clinical-grade.

If a trend creates urgent concern, skip the chatbot and contact an appropriate healthcare professional or emergency service. Appointment preparation is not triage.

Make a privacy-minimized note

Use a table like this before opening Gemini:

Trend | 90-day observation | Relevant context | Question to organize

Sleep duration | Weekly average moved from about 7:00 to 6:20 | Two weeks of travel; four missing nights | Is the change large enough to discuss or monitor?

Resting heart rate | Four-week average is 6 bpm above the first month | New exercise plan began mid-period | What context would help interpret the change?

Steps | Weekday average fell about 18 percent | Work schedule changed | Is this relevant to the visit, and what target is appropriate?

Remove name, date of birth, address, precise location, medical-record number, and other identifiers. Use approximate weeks instead of exact dates unless the date is medically relevant. Do not include a medication list, diagnosis, or laboratory result unless the clinician has asked for it and the sharing method is appropriate.

The note is still health information. Review Gemini's account activity and privacy settings before using it, and do not use an employer-managed or shared account for personal health details without understanding the policy.

The checkup-prep prompt

Paste only the minimized table and use this instruction:

Help me organize questions for a routine checkup. Do not diagnose, interpret
these trends as a medical condition, recommend medication changes, or set a
treatment plan.

For each observation, identify what context is missing and draft one neutral
question I can ask a qualified healthcare professional. Then create a one-page
brief with: the observation, relevant context, missing information, and the
question. Keep uncertainty explicit.

[paste the three-row table]

The boundaries matter. Asking "What disease do these numbers show?" invites a conclusion the data and tool are not qualified to make. Asking "What context is missing, and how can this question be phrased clearly?" uses the model for organization.

Read the result closely. Remove any diagnosis, causal claim, or urgent-sounding conclusion that Gemini added. A model may overstate a small change because it lacks the full clinical picture, device accuracy, baseline, and medical history.

Turn the output into a one-page appointment brief

The final brief should fit on one page:

1. Reason for the visit: routine checkup or the clinician's stated purpose
2. Three observations: plain numbers, time range, and tracking gaps
3. Relevant context: travel, schedule, device changes, or other facts
4. Three questions: one per observation
5. Clinician notes: blank space for the answer and next step

Avoid a long AI-generated explanation. The brief is meant to improve the conversation, not consume the appointment.

Good questions are neutral and specific:

• "Is this change meaningful enough to monitor, given the missing tracking days?"
• "What additional information would help interpret this trend?"
• "Should this be compared with a clinical measurement or a longer baseline?"
• "Are there symptoms or changes that should prompt an earlier follow-up?"

Only a qualified professional should answer those questions in the context of the person, history, and examination.

Check the device and data limitations

Wearable and phone measurements can have gaps. The device may have been removed, worn loosely, charged overnight, or replaced during the period. Software updates can also change presentation or calculation.

Before treating a change as real, check:

• How many days have data
• Whether the same device measured the full period
• Whether the metric definition changed
• Whether travel, illness, or schedule changes explain part of the pattern
• Whether the number is a direct measurement or an estimate

Put these limitations in the brief. Missingness is information.

A safer division of work

Use the tools for different jobs:

• Apple Health: displays and exports the personal measurements
• The person: selects relevant trends, adds context, and removes identifiers
• Gemini: organizes observations and drafts neutral questions
• A qualified healthcare professional: interprets the information, decides whether further evaluation is needed, and recommends care

This division keeps AI in an administrative role. It also makes the appointment more efficient because the clinician sees a compact record of what changed and what the person wants to understand.

Where this can go wrong

Too much data: Uploading the full archive exposes more information than the prompt needs. Use three manually selected trends.

False causation: A model connects a trend with a condition or behavior without enough evidence. Delete the claim and ask the clinician.

Missing context: A device gap looks like a health change. Add tracking coverage and device notes.

Urgency drift: The output sounds alarming or reassuring. Do not use it for triage. Seek qualified help based on symptoms and established medical guidance.

Account confusion: Personal health information goes into a shared, work-managed, or unfamiliar account. Check ownership, activity settings, and privacy terms first.

Action Steps Summary

1. Review 90 days in Apple Health and choose no more than three relevant trends.
2. Transcribe only the needed observations, add context, and remove identifiers.
3. Ask Gemini to organize neutral clinician questions, not to diagnose or recommend treatment.
4. Delete unsupported conclusions and bring a one-page brief to the appointment.
5. Let a qualified healthcare professional interpret the data and decide next steps.

This information is for educational purposes only and is not medical advice. Please consult a qualified healthcare professional before making changes to your health routine.

PromptHacker Premium helps Executives use AI for preparation and organization while keeping private data, medical interpretation, and care decisions in the right hands.

Sources

• Apple: Share health and fitness data in XML format
• Google Gemini Apps Privacy Hub
• Google Gemini Apps file upload guidance