Aged Care Cognitive and Mood Screening Records: Use case and AI Confidentiality

When a new resident seems vague, or a settled one stops eating and stops joining in, a nurse screens thinking and mood with tools such as the RUDAS and the Cornell Scale for Depression in Dementia, then has to turn the scores and observations into a record the doctor can use. That is a natural job to give to ChatGPT or Claude. The record joins a named resident to a score, what he said about himself, how he behaves in the evening, who was told and what his wife asked. This page marks every item in a realistic screening record, shows which obligations bear on each one and shows how the same record is written up on Ermos inside the service.

The cognitive and mood screening record

Completed by a registered nurse after admission and again when a resident’s mood, behaviour or thinking changes, then given to the GP and filed in the resident’s record.

Kurrajong Gardens Aged Care
Cognitive and mood screening record · Banksia wing · Entry Tuesday 17 November 2026 ·

Resident
Why screened
RUDAS
Cornell
What he said
Behaviour
Diagnosis and doctor
Plan
Wife
SignedBeatriz Santos RN, 3.45pm.

An invented record. The names, scores and statements are invented and the mobile number is a placeholder. Hover or tap any highlighted item for the analysis; the full list is below.

Screening record summariser

An Agent+ that prepares the screening record from the scores and notes the nurse enters. It sets each cognitive and mood screen beside the earlier ones, notes what the resident, family or staff reported, and drafts a short summary for the GP. Interpretation stays with the clinicians.

What it does

  1. Reads the screening scores and notes recorded for the resident
  2. Sets each result beside the earlier screens
  3. Drafts a short summary for the GP with open items

What stays with your people. The registered nurse checks the record and the GP interprets it. The agent summarises; it offers no clinical opinion.

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 service

  1. 1
    The resident

    A named man with his room, date of birth and admission date. Section 168(1) of the Aged Care Act 2024 requires a registered provider to ensure the protection of personal information about the people it delivers services to, and not to use it other than for a purpose connected with delivering the service or the purpose it was given for. Information collected in providing aged care can be health information (section 6FB(3)(b) of the Privacy Act). Privacy Act guide.

  2. 2
    Why he was screened, and his wife’s report

    A wife’s account of her husband’s eating, sleep and hopelessness. Information about his health, including his psychological health (section 6FB(3)(a)), is health information under section 6FA, and it is also sensitive information under section 6(1). It is also personal information about her. Outcome 5.4 asks the provider to support early identification of, and response to, changing clinical needs (rule 15-30(11)). Patient data guide.

  3. 3
    The RUDAS score

    A 30-point screen, completed on a named day with a named nurse. A score is a prompt for further assessment and not a diagnosis, and the cut-offs used in the literature differ between studies. Outcome 5.6 asks the provider to ensure people who experience cognitive impairment, whether acute, chronic or transitory, receive comprehensive services (rule 15-30(13)). The Rules do not name a tool. A tool asked to interpret a 19 may turn a screen into a label. Patient data guide.

  4. 4
    The Cornell total and its cut-offs

    A 19-item scale scored from interviews with the resident and people who know him. The developers’ thresholds are above 10 for probable major depression and above 18 for definite, and it is a screen rather than an instrument for diagnosis. The interviews also record what his wife and a worker said. A tool can state a threshold wrongly or rename a screening result as a diagnosis, which only a doctor can make. Patient data guide.

  5. 5
    What he said about himself

    The most sensitive line in the record. It is his own words about wanting not to wake, with the answer to a direct question and the clinical response. Rule 15-20(7) says risks to individuals, and changes and deterioration in their condition, are escalated and communicated as appropriate, and Outcome 5.4 asks for coordinated clinical care (rule 15-30(10)). The response is the nurse’s and the doctor’s. A tool must not soften it, shorten it or add a risk rating of its own. Patient data guide.

  6. 6
    Behaviours and a possible cause

    Evening calling out, refused showers and a shoulder that may be sore. Outcome 5.6 asks the provider to identify situations and events that may lead to changes in behaviours (rule 15-30(14)). The note is a nurse’s working observation. A tool asked to summarise it may label the behaviour or lose the possible pain. Patient data guide.

  7. 7
    No diagnosis, a named doctor and a clinic

    That no diagnosis has been made, the doctor who will review him and a hospital clinic. A named GP and hospital tie a health service to him. Whether to start any medicine is for Dr Nair, Mr Gleeson and his wife, and a tool that writes “depression” or “dementia” in the diagnosis line has put something in the record that nobody has decided. Patient data guide.

  8. 8
    The plan and the check interval

    Rule 15-20(2)(b) says care and services plans must include strategies for risk management and preventative care. The 30-minute checks and the repeat date are clinical decisions for him, and the activities are the lifestyle team’s. A tool can change an interval or a date. Patient data guide.

  9. 9
    His wife, her role and her request

    A named wife, her number, her legal role, his agreement that she may have the results and her request about their son. Rule 15-15(13) requires the information of individuals to be kept confidential and managed appropriately, in line with their informed consent. His agreement to share with her is not agreement to share with a tool, or with their son. Privacy Act guide.

  10. 10
    The names of the two tools

    The names of the RUDAS and the Cornell Scale, their length and scoring range. Published, with nothing about a resident in them.

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 service 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 resident’s 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 aged care providers.

Common questions

Do ChatGPT and Claude use the information in a cognitive and mood screening record?
Yes, a pasted record is sent to the AI company and processed there to produce the 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 record joins a named resident to a RUDAS and a Cornell score, his own words about not wanting to wake and his wife's number and request.
Can we use ChatGPT or Claude for a screening record if we remove the resident's name?
No, because a screening record without a name still describes an identifiable man. 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. His date of birth, his room, the Monday he was admitted and his wife's account of his sleep and eating point to one person. 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.
Can ChatGPT or Claude remember the details in a screening record later?
Both products also have memory features that can carry details from one conversation into later ones, and on a personal account each user decides whether they are switched on (ChatGPT, Claude). That matters here because the record holds a resident's mood scores, his words about himself and his wife's request in a form that identifies him. A home should decide whether staff may use these products with resident records at all.
How can an aged care home use AI for a cognitive and mood screening record 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 nurse asks for a write-up the doctor can use. It is built from the resident's record and the home's own screening format, and a registered nurse checks the scores and signs it.
What does the aged care law say about cognitive impairment?
Outcome 5.6 of the strengthened Aged Care Quality Standards, in rule 15-30 of the Aged Care Rules 2025, says a provider must ensure people who experience cognitive impairment, whether acute, chronic or transitory, receive comprehensive services that optimise clinical outcomes and match their needs, goals and preferences. It must also identify situations and events that may lead to changes in behaviours. The Rules do not name a screening tool.
Is a screening score a diagnosis?
No. The RUDAS is a 30-point screen and the Cornell Scale is a 19-item scale scored from interviews. The published cut-offs prompt further assessment and are not a diagnosis, and RUDAS cut-offs differ between studies. A doctor makes any diagnosis. A tool that turns a score into a label goes beyond what the instrument supports, so a nurse should check what it writes.
Is information about mood and cognition sensitive information?
Where it is health information, yes. Section 6FA of the Privacy Act 1988 covers information or an opinion about a person's health, and section 6FB(3) treats an activity in the course of providing aged care that assesses or manages a person's health, including psychological health, as a health service. Health information is sensitive information under section 6(1), which brings stricter rules in the Australian Privacy Principles.

See Dominion answering from your own documents.

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

Book a demo