Healthcare Outcome Measures Tracking Spreadsheets: Use case and AI Confidentiality

An allied health clinic keeps one sheet of the questionnaire and functional scores it collects from each client, so it can see who is improving and who is not. Before the clinic review the practice manager pastes the whole sheet into ChatGPT or Claude to chart the trends, spot the clients who have stalled and write an outcomes summary. The sheet names every client beside a K10, a PHQ-9, a knee score or a neck pain score, and the notes beside them say what is going on at home. This page marks each item in an invented outcome measures spreadsheet that the Privacy Act and NSW health records law bear on, and shows how the same sheet is analysed on Ermos inside the clinic.

The outcome measures tracking spreadsheet

Kept by the practice manager and clinicians and updated after each reassessment, then reviewed before the clinic’s service review meeting. Each row ties a named client to a service the clinic provides.

fx=E2-D2
ABCDEFGH
1ClientClinicianMeasureBaselineLatestChangeLatest dateNote
2K1028-616 Nov
3Oxford Knee Score14126 Nov
4Meera JoshiConners inattention T783 Nov
5Callum ReidPSFS average3.01.713 Oct
6Meera JoshiPHQ-919-812 Nov
7Callum ReidLEFS31219 NovHip replacement in September. On track
8Meera JoshiK1031216 Nov
9Callum ReidNDI34-1210 NovDesk worker with neck pain. Back at work full days from 30 Nov
10Meera JoshiK1027-617 Nov
11Callum ReidLEFS44110 Nov
12Meera JoshiPHQ-914-816 NovGoal met. Two sessions left
13
Outcome measuresBaseline onlySummary

An invented sheet. The clinic, its staff and every client are fictional and all scores are invented. Hover or tap any highlighted cell for the analysis; the full list is below.

Outcome measures sheet updater

An Agent+ that updates the clinic's outcome measures sheet from reassessment forms and notes, keeping the practice's own columns. It compares baseline and latest scores, notes where a score has moved, and flags rows where a risk note or a missing measure needs a clinician's eye.

What it does

  1. Reads new reassessment forms against the existing sheet
  2. Updates baseline and latest scores in the clinic's layout
  3. Flags missing measures and rows needing clinical review

What stays with your people. A clinician reviews every flagged row and decides what the scores mean. The practice manager checks the sheet before the service review meeting.

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
    Client names and ages

    Each row ties a named person to a service the clinic provides, which is health information (Privacy Act s 6FA(a)(iii)) and so sensitive information under s 6(1). A business that provides a health service and holds health information is not a small business operator (s 6D(4)(b)). The OAIC recommends as best practice that organisations do not enter personal information, particularly sensitive information, into publicly available generative AI tools. Keeping scores to guide care and to review the service is one thing the clinic can explain to clients. Sending the sheet to an outside tool is a disclosure to someone else, and whether clients would reasonably expect it is for the clinic to answer (APP 6.2(a)(i); HPP 11(1)(b)). Patient data guide.

  2. 2
    Baseline and latest scores

    Questionnaire and functional scores recorded about a named person are health information (s 6FA(a)(i) and (b)). Which direction is better differs by measure: a lower K10 or PHQ-9 total is an improvement, and a higher Oxford Knee Score, LEFS or PSFS is. HPP 9 asks an organisation not to use health information without reasonable steps to ensure it is relevant, accurate, up to date, complete and not misleading, and APP 10.2 is to the same effect for use or disclosure. A chart or ranking an AI tool produces needs a person to check the direction of each scale. Patient data guide.

  3. 3
    A risk note and a safety plan

    A note recording passive thoughts and a safety plan beside a K10 score. It is sensitive information, and Code of conduct for psychologists 8.5(a) and (b) ask for accurate, factual records held securely. Clients can ask for access to their records, and APP 12.3(a) allows an organisation to refuse where access would pose a serious threat to the life, health or safety of any individual. Sorting or summarising the sheet can drop a clinical flag, and Ahpra says practitioners must apply human judgment to any output of AI. Patient data guide.

  4. 4
    Labels such as non-responder

    These are opinions about identifiable people, and the Privacy Act treats an opinion as personal information whether or not it is true (s 6(1)). Code 8.5(c) for psychologists and section 8.3 of the Shared Code of conduct ask that records show respect and avoid demeaning or derogatory remarks. The OAIC says AI output that contains personal information is also covered, so a ranking of who is least likely to improve is new information about each person on it. Privacy Act guide.

  5. 5
    A young person and her mother

    Under the NSW Act a private sector health service provider keeps health information collected while the individual was under 18 until she turns 25 (s 25(1)). Code 4.2(d) asks psychologists to act according to a client’s capacity for decision-making and consent, including young people, and to consider the need for the consent of a parent, carer or guardian. The row shows who consented and who attends. Patient data guide.

  6. 6
    Family and personal circumstances

    A wife’s illness, a daughter’s wedding, a baby, a partner and an employer who has not been told. They are personal information about the client and about people who are not clients, and a wife’s Parkinson’s is health information about her (s 6FA(a)(i)). They explain the score, but they are in the sheet so the clinic can care for the client, and no one agreed they would go anywhere else. Privacy Act guide.

  7. 7
    A discharge and why

    The row records that a named client is to be discharged after three missed sessions. Section 4.7 of the Shared Code of conduct says good practice involves informing the patient adequately of a decision to end the relationship and facilitating arrangements for continuing care. The sheet, the letter to the client and the clinical record should say the same thing.

  8. 8
    Clinician names and who sees whom

    Employee names and caseloads. Under section 8(1)(a) of the Privacy Act an act done by a person employed by an organisation in the performance of their duties is treated as the organisation’s act, so a staff member pasting the sheet into a chatbot is the clinic’s act, and may be a disclosure by the clinic.

  9. 9
    The summary row

    Counts of rows and outcomes with no client named: 11 rows, 9 improved, 1 worse and 1 baseline only, which a person can check against the rows. It carries no personal information. The clinic review date and the name of the person who compiled it are low risk alone.

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 an outcome measures spreadsheet?
Everything pasted or uploaded into ChatGPT or Claude 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. An outcome measures spreadsheet puts eleven named clients beside their K10, PHQ-9 and knee scores, a risk note with a safety plan, and a sixteen-year-old with her mother.
Can we use ChatGPT or Claude for an outcome measures spreadsheet if we remove the names?
Not safely. 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. The notes beside the scores describe a wife with Parkinson's, a daughter's wedding and a baby five months old, which can point to one client, and each row also names the clinician. 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.
What travels with the outcome measures spreadsheet if we upload the file to ChatGPT or Claude?
An uploaded file leaves the business as a whole file, so the other tabs go with it, including Baseline only and Summary, along with any hidden columns, comments, tracked changes and file properties. In this workbook that means every client's baseline and latest score, the notes, the clinician names and the formula behind the change column. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. Check every tab before any upload.
How can an allied health clinic use AI for an outcome measures spreadsheet 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 trends and the outcomes summary are produced from the clinic's own sheet, and a clinician checks the direction of each scale and every risk flag before the clinic review.
Is a spreadsheet of outcome scores health information?
Generally yes. Under section 6FA of the Privacy Act, information about the health of an individual, or about a health service provided to them, is health information when it is also personal information, and so is other personal information collected in providing a health service. A row with a named client and a score ties a person to a service. Health information is sensitive information under section 6(1).
Can a client ask for their scores?
Yes, they are part of the record. APP 12.1 requires an organisation to give an individual access to the personal information it holds on request, with the exceptions in APP 12.3. In NSW a request under the Health Records and Information Privacy Act is made in writing (section 26) and a private sector person must respond within 45 days (section 27).
Does a small allied health clinic have to follow the Privacy Act?
Generally yes. Section 6D(4)(b) of the Privacy Act takes a business out of the small business exemption if it provides a health service to another individual and holds health information other than in an employee record. The NSW Health Records and Information Privacy Act 2002 also applies to private sector health service providers, and section 4 lists physiotherapy and psychology services as health services.

See Dominion answering from your own documents.

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

Book a demo