Healthcare Food Diary Analysis Spreadsheets: Use case and AI Confidentiality

A client keeps a food and symptom diary for a week, and the dietitian wants the days totalled, the patterns found and a first draft of feedback. It is a spreadsheet, so it is tempting to upload it to ChatGPT or Claude and ask for AI food diary analysis. The sheet carries more than meals. It holds the client’s name and mobile, morning weights, blood results, gut symptoms, an eating disorder history and remarks about her flatmate and her mother. This page marks each item in an invented diary that the Privacy Act, NSW health records law and the Dietitians Australia Code of Conduct bear on, and shows how the same sheet is worked on Ermos inside the practice.

The food diary analysis spreadsheet

Kept by a client between visits and sent to the dietitian before a review. The dietitian reads it, adds up the intake, notes the symptoms and decides what to change. A summary usually goes to the referring doctor.

Bellamy Nutrition
Food and symptom diary, working copy · Monday 9 to Sunday 15 November 2026 · Anouk Bellamy, APD

fx=AVERAGE(C2:C8)
ABCDEFG
1Daykcal (app estimate)Protein gMorning weight kgSymptomsNotes
2Mon 9 NovCoffee and toast with butter. Wrap with chicken and salad. Rice and vegetable stir fry.38Normal day
3Tue 10 NovPorridge with milk. Sandwich. Pasta with tomato sauce, an apple.44None
4Wed 11 NovCoffee and a banana. Soup. No dinner.31Work drinks and I was not hungry after
5Thu 12 NovYoghurt and granola. Tuna salad. Chicken, potato and green beans.40Bloated, 3/10
6Fri 13 NovEggs on toast. Burrito bowl. Two slices of pizza and a glass of wine.47None
7Sat 14 NovBlack coffee. Rice cakes. Nothing else.22
8Sun 15 NovPorridge. Roast chicken and vegetables at Mum’s. A biscuit.41None
97-day average
10Client
11History
12Bloods
13Consent
DiaryBloodsSummary

An invented diary. The client, her flatmate, her mother and the practice are fictional, the mobile number is a placeholder and every figure is made up. The totals in row 9 are the averages of rows 2 to 8. Hover or tap any highlighted item for the analysis; the full list is below.

Food diary summary builder

An Agent+ that reads a client's food diary and tabulates it in the layout your practice uses: meals by day, weights, results and symptoms. It sets the diary beside the targets in the client's plan and marks entries the dietitian may want to ask about.

What it does

  1. Reads the diary, weights and recorded results
  2. Tabulates meals and symptoms by day
  3. Flags entries to raise with the client

What stays with your people. The dietitian reads the summary, interprets it and decides what changes. The agent tabulates; it does not advise the client.

An Agent+ is built once and anyone in the practice can run it. It works from your own documents and runs on Ermos Dominion, the AI unit in your office, or on Ermos Edge, your own instance in an Australian data centre.

What leaves the practice

  1. 1
    Name, date of birth, address and mobile

    Four identifiers that tie every row to one person. A business that provides a health service and holds health information is not a small business operator under the Privacy Act (s 6D(4)(b)), so its turnover does not take it outside the Act, and the NSW Act lists dietitian services as a health service (s 4, para (j1)). APP 11.1 asks for reasonable steps to protect personal information from misuse, interference, loss and unauthorised access or disclosure. Standard 2.2.1 of the Dietitians Australia Code asks dietitians to comply with the privacy and health records laws that apply, including when they use artificial intelligence. Privacy Act guide.

  2. 2
    A past eating disorder and who knows about it

    A mental health history, the treatment she had and which people in her life know. It is information about her health at any time (s 6FA(a)(i)), and health information is sensitive information (s 6(1)). APP 3.3 asks for consent to collect sensitive information that is reasonably necessary for the practice; she gave it for her own care. APP 6.1, HPP 10(1) and HPP 11(1) allow use or disclosure for another purpose only with consent or an exception, such as a directly related purpose she would reasonably expect. The OAIC recommends as best practice that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools. Patient data guide.

  3. 3
    Blood results and the referral

    Pathology values from the referring doctor, with a date and a symptom history. Health information (s 6FA(a)(i)). The sheet carries them from Dr Eze’s referral, so passing them on turns on APP 6 and HPP 11. An AI tool can restate a value or a unit wrongly or comment on a result as if it were advice, and standard 4.1.4 of the Code asks dietitians to assess evidence sourced through generative AI before applying it. The decision not to start an elimination diet is the dietitian’s. Patient data guide.

  4. 4
    Seven morning weights

    A weight a day, written down by the client. Information about her health, collected in providing a health service (s 6FA(a)(i) and (b)). An AI tool can chart the series, call it a loss or comment on it, and how weights are discussed with a client who has this history is a clinical decision for the dietitian. HPP 9 asks that health information is relevant, accurate, up to date, complete and not misleading before it is used. Patient data guide.

  5. 5
    Daily calorie estimates

    Figures from the client’s own app, which are estimates. The OAIC says personal information includes information about an identifiable person that an AI model infers or generates, and that generating it is a collection that must comply with APP 3. APP 10.2 asks for reasonable steps to make sure information an entity uses or discloses is accurate, up to date, complete and relevant. An AI tool can total the column wrongly, and the dietitian decides whether the client sees totals at all. Privacy Act guide.

  6. 6
    Symptoms set beside the meals

    Bloating, cramps and dizziness on named days. Health information (s 6FA(a)(i)). This is the column an AI tool is asked to find patterns in, and a pattern it finds is a suggestion for the dietitian to assess. An average across the week can hide the day that matters, here Saturday 14 November, which a person has to read for themselves. Patient data guide.

  7. 7
    A flatmate, a mother and what she said

    Remarks about a named flatmate and about what her mother said and did. They are personal information about people who did not agree to be in a file or a prompt, and they are also part of the client’s own story, so APP 6.1 and HPP 11(1) govern passing them on. APP 11.1 applies to everything the practice holds, including what clients say about others. Privacy Act guide.

  8. 8
    The column headings

    Day, foods and drinks, kcal, protein, morning weight, symptoms and notes: the practice’s own blank diary template, with no client in it. Nothing to protect by itself, and the same wording is in any dietetic handout.

  9. 9
    The averages and the dietitian’s note

    A plain AVERAGE formula, three results and a note to herself. The arithmetic is harmless, but the average of 1,160 kcal hides a day of 760, and the sheet it sits in holds everything else. Standard 3.1.3 of the Code asks dietitians to follow Ahpra’s guidance on AI, which says practitioners must apply human judgment to any output of AI, and standard 4.1.4 asks them to assess the evidence before applying it.

  10. 10
    Consent and the missing mention of AI

    Eliza agreed to the diary and to a summary going to her doctor, and AI use was not discussed. The Dietitians Australia Code asks dietitians to fully explain the proposed activity, including any use of artificial intelligence (3.2.1), to obtain informed consent or other valid authority before transmitting or sharing a client’s information (3.2.2(iv)) and to respect a client’s right to withhold or withdraw consent (3.2.3). Ahpra’s guidance, which standard 3.1.3 tells dietitians to follow, says to inform patients about AI use and ideally note the response in the record. The Code, approved by the Dietitians Australia Board in June 2026, applies to Dietitians Australia members and to APDs. Dietitians are not among the professions Ahpra regulates, so it is a professional standard. Patient data guide.

What happens in Frontier Models (ChatGPT, Claude)

  • The whole document is transmitted to the provider’s servers. On personal plans the user cannot choose where it is stored or processed, and for the major providers that has generally meant outside Australia, under that country’s legal process. Some business plans offer storage in Australia, but the provider still receives the content.
  • On consumer tiers, prompts can be retained and used to train future models.
  • The provider’s staff and systems can reach the content under the provider’s own policies.
  • Once sent, the practice no longer controls it. Retention and deletion happen on the provider’s terms.

The same document on Ermos

Ermos Dominion

The document is drafted on a unit in your office, from your own templates and the patient record. Every marked item stays on the unit, which keeps working with the internet switched off. Nothing in it is transmitted to an AI provider, so there is no disclosure to analyse: the confidentiality work is the same as for your file server.

Ermos Edge

Your documents, the index and your access boundaries stay on your own instance in an Australian data centre. When a staff member asks for this draft, only their question and the excerpts they are permitted to see go to a named Australian provider, who is contractually barred from keeping the content or training any model on it. Nothing is processed outside Australia.

Frontier Models (ChatGPT, Claude)Ermos EdgeErmos Dominion
Where the document is processedProvider servers, generally offshore on personal plansYour instance in an Australian data centre; answers generated in AustraliaThe unit in your office
Who can read itProvider staff and systems, under policyYour staff, within the boundaries you set. The named Australian provider processes only permitted excerpts, under contract, and keeps nothingYour staff only
Is it keptOn the provider’s termsKept on your instance; the provider is barred from retaining itStays on the unit
Can it train a modelOn consumer tiers, yesNo, by contractOnly your own unit learns from it, on your hardware

Part of Use cases in healthcare practices. See how this works in practice: Clinical documentation on Ermos.

Common questions

Do ChatGPT and Claude use the information in a food diary analysis spreadsheet?
Yes, an uploaded or pasted diary is sent to the AI company's servers and processed there to produce a reply. As at October 2026, whether a conversation on a personal plan (ChatGPT Free, Plus or Pro; Claude Free, Pro or Max) can be used to train the models depends on a setting each user controls. Business plans (ChatGPT Business and Enterprise, Claude Team and Enterprise) and the developer APIs sit under commercial terms, and OpenAI and Anthropic each say they do not train on that content by default (OpenAI, Anthropic). On either product, the thumbs up and thumbs down buttons can send a whole conversation back for training. This diary holds seven morning weights, a ferritin result, cramps and dizziness on named days, and a past eating disorder.
Can we use ChatGPT or Claude for a food diary if we remove the client's name?
Not safely, because the diary still describes one woman. Removing names rarely makes a document anonymous. Information stays personal information while the person is reasonably identifiable, and the OAIC warns that information can be at risk of re-identification even when it has been de-identified or anonymised. Her mother, her flatmate Mel, her ferritin and haemoglobin results and an anorexia nervosa history that her flatmate does not know about would let anyone who knows her pick her out. Her health details are sensitive information under the Privacy Act. The Office of the Australian Information Commissioner (OAIC) recommends in its guidance on commercially available AI products that organisations do not enter personal information, and particularly sensitive information, into publicly available generative AI tools, because of the significant and complex privacy risks.
Does uploading a food diary spreadsheet to ChatGPT or Claude send more than the visible cells?
It can, because an uploaded file leaves the practice as a whole file. A spreadsheet can carry other sheets, hidden columns, comments, tracked changes and file properties. This workbook has Diary, Bloods and Summary tabs, so a file uploaded for the Diary tab would carry the other two as well. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product.
How can a dietitian use AI for a food diary analysis without sending it to ChatGPT or Claude?
Ermos is built so the document stays under your control. On the Dominion AI unit it is drafted inside your office and none of it is sent to an AI provider. On Ermos Edge your documents and index stay on your own instance in an Australian data centre, and only the question and the excerpts the user is permitted to see go to a named Australian provider that is contractually barred from keeping them or training any model on them. The totals and a first draft of feedback are produced from the client's diary and the practice's own diary template. The dietitian checks every figure and decides what the client sees.
Can a dietitian upload a client's food diary to ChatGPT or Claude?
Standard 2.2.1 of the Dietitians Australia Code of Conduct asks dietitians to comply with privacy and health records law when using artificial intelligence, 3.2.1 asks them to explain any use of AI, and 3.2.2 asks for informed consent before sharing a client's information. The OAIC recommends as best practice that organisations do not enter personal information, particularly sensitive information, into publicly available generative AI tools.
Is a food diary health information under the Privacy Act?
Usually. Section 6FA(b) covers other personal information collected in providing a health service, and health information is sensitive information under section 6(1). A business that provides a health service and holds health information is not a small business operator under section 6D(4)(b). The NSW Health Records and Information Privacy Act 2002 also lists dietitian services as a health service in section 4. Weights, symptoms and blood results in the same sheet are plainly health information.
Does the dietitian still have to check numbers an AI adds up?
Yes. Standard 4.1.4 of the Code asks dietitians to assess evidence sourced through generative AI before applying it, and Ahpra's guidance, which standard 3.1.3 tells dietitians to follow, says practitioners must apply human judgment to any output of AI. A calorie total built from a client's own app estimates is an estimate of an estimate, and the dietitian decides how and whether to share totals.

See Dominion answering from your own documents.

A demonstration takes 30 minutes, uses no client data, and comes with no obligation.

Book a demo