Gains:
- Ability to prepare a complete intake interview guide and structured nutrition history with artificial intelligence
- Ability to verify portion estimates with the client and measurement while converting free narratives, such as a 24-hour nutritional reminder, into a table
- Ability to conduct the interview and trust relationship as a dietitian and position artificial intelligence only in the preparation and editing phase.
Every successful nutrition plan starts with a good "intake." Intake is the process of systematically collecting the client's lifestyle, health history, eating habits, goals and restrictions at the first meeting. Incomplete or incorrectly gathered information at this stage disrupts every subsequent step: wrong story, wrong calculation; Wrong calculation leads to wrong plan. Artificial intelligence plays two powerful roles at this stage: first, preparing a comprehensive and consistent interview structure; Second, to turn the scattered notes collected into an organized story. But remember — the AI doesn't conduct the interview, it prepares the interview. The relationship of trust and clinical intuition established with the client belongs to the dietitian.
Components of a good purchase
A comprehensive nutritional history includes the following topics: personal information (age, gender, occupation, activity level), anthropometric measurements (height, weight, waist circumference; "anthropometry" is the science of body measurements), medical history (chronic diseases, surgeries, family history), medication and supplement use, allergies and intolerances, laboratory results, 24-hour nutritional reminder (the client explains step by step what he ate yesterday), eating behavior (meal skipping, emotional eating, frequency of eating out), liquid and alcohol consumption, sleep and stress, goals and motivation. AI is very good at producing an interview guide that doesn't skip any of these topics; Human memory can skip a title at the end of a tiring day.
The following table shows the type of ingestion data and the AI's relationship to it:
Data type
example
Role of AI
Dietician responsibility
objective measurement
Height, weight, analysis
Editing, account draft
confirm the accuracy
subjective narrative
"I get very hungry in the evenings"
Summarizing themes
Clinical comment, follow-up question
Security-critical
allergy, medication
Checklist reminder
peer-to-peer verification
motivational
target, obstacle
Question suggestion
Empathy, relationship building
Step by step: Preparing for a hiring interview with AI
- Produce an interview guide. Ask the AI for a list of questions with complete headings appropriate to the client profile. For example, a pregnant client requires additional items (nausea, iron, folic acid).
- Simplify the questions. Have it translated into everyday language that the client will understand. “How your blood sugar behaves after a meal” rather than “postprandial glycemic response.”
- Do the interview yourself. Don't let the AI into the interview room; Use the guide as a guide, establish eye contact and trust with the client.
- Have the notes edited by AI. After the interview, give your scattered notes (anonymised) to AI and have them converted into a structured story text.
- Identify the deficiencies. Ask the AI “what critical information is missing from this story?” ask; Prepare a list of questions for the next interview.
Tip: Use the AI-generated list of questions not to "read it as is" but to "make sure you don't miss any topics." The client's narrative should determine the flow of the conversation, not a rigid survey.
Balancing open-ended and closed-ended questions
A good recruitment interview balances two types of questions. Closed-ended questions get short, precise answers ("How many meals do you eat?", "Do you consume dairy?") and are ideal for collecting objective data. Open-ended questions invite the client to explain ("Can you describe your day in terms of food?", "What is your relationship with food?") and often reveal the most valuable clinical clues, emotional eating and real obstacles. Oftentimes, beginners create a survey with only closed-ended questions and miss the client's story. Explicitly demand this balance when asking the AI for a list of questions: “have at least one open-ended exploratory question in each topic”. This way, the interview becomes a real discovery, not a form-filling session. You can also have the AI prepare “follow-up question” suggestions (a deepening question to ask after an answer); For example, when the client says "I snack in the evenings", the question to be asked is "when does this usually happen and with what emotion?" like.
Configuring 24-hour food reminder
The most valuable but messiest part of the reception is the 24-hour reminder. The client explains, "I drank tea in the morning, some cheese, kebabs outside at noon, and pasta at home in the evening." It takes time to put this free text into a table by meal by meal, estimated portion and food group. AI accelerates this transformation: it turns the free narrative into a structured picture. But portion estimates are rough; "some cheese" could be 30 grams or 80 grams. AI may suggest a range, but you clarify the actual portion by asking the client.
three mini cases
Case 1 — Catching the missing title. A dietitian conducts a 40-minute interview with a 52-year-old male client, gives his notes to the AI and asks, "Is there a missing title?" he asks. AI makes you realize that medication use is never asked. In the second interview, it is revealed that the client is taking warfarin (blood thinner) — this is safety-critical information that directly concerns vitamin K-rich green leaves (spinach, broccoli). The AI prevented a possible error by recalling a title; The decision was made by the dietitian.
Case 2 — Portion trap. Another dietitian gives the client's statement "2 handfuls of nuts a day" to AI; The AI assumes this is "60 grams" and adds ~360 kcal per day. However, the client's palm is large and the actual amount is ~90 grams (~540 kcal). The 180 kcal difference means a difference of ~1,260 kcal per week. The dietitian corrects the portion by having the client measure it with a kitchen scale. Lesson: AI's assumption of portions is initialization, not measurement.
Case 3 — Anonymization. An intern uploads the transcript of the interview recording to the AI with the client's name. The supervisor intervenes: the text is full of name, workplace and neighborhood information. Together they anonymize the text and reproduce it with only clinical and numerical content. Reception data is organized, confidentiality is protected.
Copiable prompt templates
INTERVIEW GUIDE TEMPLATEYour role: assistant preparing recruitment interview for dietitian. If you don't have the interview, you won't do it, you will produce a guide. Client profile (anonymous): [age, gender, main target, known status]. Give me a list of questions suitable for this profile, covering the following topics COMPLETELY: medical history, medication/supplement, allergy/intolerance, 24-hour reminder, eating behavior, fluid, sleep/stress, goal. Write the questions in plain language that the client will understand.
TEMPLATE FOR TURNING A NOTE INTO A STORY Below are the scattered, ANONYMOUS notes of a recruitment interview. Turn them into an organized nutrition story by headings. ADDING made-up information; Do not assume anything that is not in the note. Mark unclear areas as "[to be asked to the client]". Notes: [...]
24-HOUR REMINDER CHART TEMPLATE Turn this free narrative into a meal-by-meal chart: columns =Meal | Nutrition | Estimated servings (give range) | Food group. Counting portions EXACTLY; If it is unclear, give a space and write "[confirm with client]". Narrative: [...]
MISSING DATA SCREEN TEMPLATEReview the anonymous nutrition history below. List any critical information that is MISSING (especially medications, allergies, tests, activities) for a safe and accurate plan. For each missing piece, suggest a clear question to ask at the next meeting. Story: [...]
Weak prompt / Strong prompt
Weak prompt:
Prepare questions for the client interview.
Profile, title scope and language level are unclear; The output will be superficial and general, critical headings (medication, allergies) may be skipped.
Powerful prompt:
Your role: assistant preparing intake interview guide. Client (anonymous): 29-year-old female, newly converted vegetarian, history of iron deficiency, target energy increase. Complete the following topics: medical history, medications/supplements, allergies, 24-hour reminder, B12/iron/protein sources, eating behavior, target. Write the questions in plain language. 2-3 questions are enough for each topic.
The second prompt profile gives the clinical focus (B12, iron, protein, critical in vegetarianism) and language level; The output is a directly usable guide.
Common mistakes
- Trying to have the AI do the interview: AI prepares, the human does the interview. Trust relationships do not become automatic.
- Mistaking portion assumption for measurement: AI's estimate is range; the actual portion comes from the client/measurement.
- Skipping the critical title: Not asking about medications and allergies; these affect every subsequent step.
- Uploading raw data without anonymization: Sharing the name, workplace, analysis PDF as it is is a violation of KVKK.
- Not noticing fabrication: Overlooking information that the AI “added” when editing the note but that the client did not say.
In summary
The intake phase is the foundation of the plan, and AI is the perfect preparation assistant here: producing complete interview guides, turning loose notes into organized narrative, recalling missing critical information, charting free narrative. But you do the interview, you confirm the portion, you confirm the safety-critical information (medication, allergies) directly. AI supports your memory, it does not replace your clinical intuition. And you do all this by ensuring that the data is anonymized.
Application task
Identify a hypothetical client profile (e.g. “pregnant, 2nd trimester, complaining of nausea”). Adapt the "interview guide template" to this profile and get a list of questions from the AI. Then, write a messy interview note of 5-6 sentences with your own hand and have it turned into a regular story with the "note-to-story template". Finally, run the “missing data scan” and note which critical headers the AI finds missing.
checklist
- [ ] The interview guide covers all critical topics (medication, allergies, tests).
- [ ] I conducted the interview; The AI just did the preparation and editing.
- [ ] I had the portion estimates verified by the client/measurer.
- [ ] I checked that it wasn't made up information that the AI "added" to the story.
- [ ] I anonymized all data (name, workplace, assay ID removed).
- [ ] I prepared follow-up interview questions for missing critical information.