Healthcare Post-Discharge Follow-Up Records: Use case and AI Confidentiality

When a patient comes home from hospital, the practice nurse opens a follow-up record. It brings together the discharge summary the hospital sent, what the patient says on the phone and the appointment the GP has set aside. Nurses and managers paste it into ChatGPT or Claude to tidy the call notes or turn the hospital’s requests into a task list. The record names the patient, the diagnosis, every medicine that changed, what the My Health Record showed and a carer’s circumstances. This page marks each item in a routine post-discharge follow-up record that the Privacy Act, the My Health Records Act 2012 and Good medical practice bear on, and shows how the same record is built on Ermos inside the practice.

The post-discharge follow-up record

Opened by the practice nurse when a hospital discharge summary arrives for a patient and kept open until the GP review is done. It links the hospital’s requests, the nurse’s call and the booking.

Kestrel Lane Medical Practice
Post-discharge follow-up record · Opened Monday 9 November 2026 · Updated Wednesday 11 November 2026

Patient
Admission
Source of summary
Medicine changes
Hospital requests
Phone call
Safety-net advice
Plan
Signed

An invented follow-up record. The patient and his wife are fictional and the Medicare number is a placeholder that fails its check digit. Hover or tap any highlighted item for the analysis; the full list is below.

Discharge follow-up record builder

An Agent+ that opens the post-discharge follow-up record when a discharge summary arrives. It extracts the diagnosis, medication changes and the hospital's requests, sets them in the practice's follow-up template, and lists what the nurse should raise on the call, including any carer needs noted.

What it does

  1. Reads the discharge summary and the patient's file
  2. Extracts medication changes and the hospital's requests
  3. Drafts the nurse's call list and booking note

What stays with your people. The practice nurse makes the call and the GP decides on any medication change or review. The agent prepares the record; it does not book.

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 Medicare number

    Four identifiers that tie the record to one man. The Medicare number is a government related identifier, and APP 9.2 limits when an organisation may use or disclose one. All of it is personal information that the practice holds as a health service provider. Privacy Act guide.

  2. 2
    The admission and diagnosis

    A pneumonia, a new heart rhythm problem and a heart scan finding against a named patient. Information or an opinion about a person’s health, including an illness, is health information (Privacy Act s 6FA), and health information is sensitive information under s 6(1). Patient data guide.

  3. 3
    Where the summary came from

    The record notes that the summary was read in the My Health Record. Section 61(1) of the My Health Records Act 2012 authorises a participant to collect, use and disclose health information in a registered healthcare recipient’s record for the purpose of providing healthcare to that person, in accordance with the access controls. Section 59(2) prohibits use or disclosure of information obtained from the system that Division 2 does not authorise, where the person knows or is reckless as to that. Whether pasting it into an outside tool is for providing healthcare is a question the practice should be able to answer before it happens. Patient data guide.

  4. 4
    Medicine changes

    A new blood thinner, a heart rate medicine and a stopped anti-inflammatory, with doses. A medicine list shows what a person is being treated for. It belongs to the record Good medical practice (10.5) asks a doctor to keep accurate and to hold securely against unauthorised access. Patient data guide.

  5. 5
    The hospital’s requests

    Tasks the hospital asked the GP to do and the dates for follow-up. They carry no identifier alone, but they describe one patient’s care, and a wrong date or a missed test matters more than the wording.

  6. 6
    The call and a carer’s circumstances

    What the patient said about his symptoms, together with his wife’s dementia and the family arrangements. His wife’s health is health information about a third party. Section 16B(1A) of the Privacy Act describes a permitted health situation where a practice collects a third party’s health information as part of a patient’s family, social or medical history because it is necessary to provide the service and collects it from the patient. Using it for another purpose engages APP 6.1. Patient data guide.

  7. 7
    Standard safety-net advice

    The advice a nurse gives any patient starting a new heart rhythm medicine: when to call Triple Zero and when to ring the practice. It names no one and appears in patient handouts.

  8. 8
    The plan for the GP review

    Who will see the patient, when, and the two decisions left to the doctor. It shows the patient’s schedule and that his wife will be in the waiting room. The dose decisions are clinical judgement that no draft can make.

  9. 9
    The nurse and the doctor

    Names and times for the nurse who made the record and the doctor who signed it. It shows who to ask and who is accountable for the content.

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: Correspondence summaries on Ermos.

Common questions

Do ChatGPT and Claude use the information in a post-discharge follow-up record?
Everything pasted or uploaded 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 record carries Mr Ashworth's Medicare number, a pneumonia and new atrial fibrillation, the medicines that changed, and the fact that his wife Lorraine has early dementia and he is her main carer.
Is it safe to paste a post-discharge follow-up record into ChatGPT or Claude?
No, because the record describes one man's first week home in detail. 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. The admission is health information under section 6FA of the Privacy Act. The record also notes that the summary was read in My Health Record, and section 59(2) of the My Health Records Act 2012 prohibits use or disclosure of information obtained from the system that Division 2 does not authorise, where the person knows or is reckless as to that.
Can ChatGPT or Claude remember the details in a post-discharge follow-up record later?
Possibly, depending on the account's settings. 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). Details such as the new blood thinner or Lorraine's dementia could travel into later conversations on that account, and the practice would not be able to see where they went.
How can a medical practice use AI for a post-discharge follow-up 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 record is drafted from the hospital's discharge summary and the practice's own follow-up template. The practice nurse checks every date and medicine, and the GP reviews and signs it.
Can a practice use information from My Health Record to draft a follow-up record?
Section 61(1) of the My Health Records Act 2012 authorises a participant to collect, use and disclose health information in a patient's My Health Record for the purpose of providing healthcare to that patient, following the access controls the patient has set. Section 59(2) prohibits use or disclosure of information obtained through the My Health Record system that Division 2 does not authorise, where the person knows or is reckless about it. A practice should be able to say why each use is for providing care.
Can a patient's record include health details about his wife?
Section 16B(1A) of the Privacy Act describes a permitted health situation where a practice collects a third party's health information as part of a patient's family, social or medical history, because it is necessary to provide the service and the information comes from the patient or, if the patient cannot give it, a responsible person. Using that information for a different purpose, such as an outside AI tool, falls to APP 6.
Is a small GP practice bound by the Privacy Act when it follows up a hospital discharge?
Generally yes. Section 6D(4)(b) 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. A practice that holds a patient's discharge summary is bound by the Australian Privacy Principles, including APP 6 on use and disclosure and APP 11.1 on protecting information from unauthorised access or disclosure.

See Dominion answering from your own documents.

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

Book a demo