Healthcare Missed Appointment Logs: Use case and AI Confidentiality

Every practice loses appointments to people who do not turn up, and the practice manager usually keeps a log of who missed, why they said they missed and what the practice did next. Before the practice meeting she pastes four weeks of it into ChatGPT or Claude to find patterns and to draft a missed appointment policy and some patient letters. The log names each patient with the appointment they missed, so it also shows a bereavement, a hospital admission, a heel ulcer, a patient who could not talk because her partner had arrived and a voicemail left in error. This page marks each item in a routine log that the Privacy Act and NSW health records law bear on, and shows how the same sheet is handled on Ermos.

The missed appointment log

Kept by a practice manager or reception team and reviewed before a practice meeting, usually to see who misses appointments, how often and what to do about it. Each row ties a named patient to a service the practice provides.

fx=COUNTIF(E2:E13,"DNA")
ABCDEFGH
1DatePatientAppointmentClinicianOutcomeReason recordedContactAction
2Tue 20 OctDr VossDNA
3Wed 21 OctDr EzeDNARebook within a week
4Fri 23 OctDNA
5Tue 27 OctDr EzeDNANo answer to two calls
6Thu 29 OctDr MirchandaniDNA
7Mon 2 NovSam Kovacs RNLate cancel
8Tue 3 NovDr EzeLate cancelClosed
9Tue 3 NovCallum ReidDNA
10Wed 4 NovDr VossDNA
11Fri 6 NovDr VossLate cancelRebook after 9 Nov. Nothing to employer
12Thu 12 NovSam Kovacs RNCancelledRebook for the week of 23 November
13Fri 13 NovDr VossDNANo answer
14
Missed appointmentsBy clinicianNotes

An invented log. The patients and staff are fictional and the phone numbers are placeholders (mobiles are 04xx xxx NNN). Hover or tap any highlighted cell for the analysis; the full list is below.

Missed appointment pattern reviewer

An Agent+ that reads the practice's missed appointment log and prepares a short summary for the practice meeting. It groups entries by pattern, service and follow-up status, and marks rows where a clinical concern may sit behind the miss. It follows the practice's own recall policy.

What it does

  1. Reads the log and the practice's recall policy
  2. Groups entries by pattern, service and follow-up status
  3. Flags rows that need a clinician's attention

What stays with your people. The practice manager checks the summary, and a clinician decides on any patient follow-up. The agent summarises; it does not contact patients.

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
    Patient names, ages and the appointment missed

    Each row ties a named person to a health service provided or to be provided, which is health information (Privacy Act s 6FA(a)(iii); NSW HRIP Act s 6(a)(iii)), and health information is sensitive information under s 6(1). The appointment type alone can say a lot: an implant removal, a changed mole, an ulcer, a pain medicine review. When staff paste the log into ChatGPT or Claude the practice may be disclosing it to the tool’s owner, so APP 6.1 and HPP 11(1) apply, and the OAIC recommends as best practice not entering personal information, particularly sensitive information, into publicly available generative AI tools. Patient data guide.

  2. 2
    The reasons patients gave

    Chest pain, a fever, a fall and a hospital admission are information about a person’s health (s 6FA(a)(i)). A bereavement, a lost job and a family member’s call are personal information about the patient and about the third party who rang (s 6(1)). The reasons were given to explain an absence, and a practice that keeps them for its own planning is a different thing from sending them to an outside tool. APP 11.1 and HPP 5(1)(c) ask for reasonable security safeguards against unauthorised access, use or disclosure. Patient data guide.

  3. 3
    A safety flag and a request to stop reminders

    A patient could not talk because her partner had arrived and asked for no more SMS reminders. In a pasted log her name, her appointment and her situation sit together. APP 11.1 and HPP 5(1)(c) ask for reasonable safeguards against unauthorised disclosure. Whether a drafted letter, message or voicemail honours the flag depends on a person checking it, because the AI only knows what the log says. Patient data guide.

  4. 4
    Phone numbers and a daughter’s call times

    Mobile numbers identify people and let someone reach them. The row for Mr Adeyemi records his daughter’s number and her call window. Under the NSW Act an attorney under an enduring power of attorney is an authorised representative (s 8(1)(a)); the representative may act for an individual who is incapable of acting (s 7(2)) but not for one who is capable unless the individual expressly authorises it (s 7(3)). The log does not say which applies to him. Privacy Act guide.

  5. 5
    Clinical follow-up instructions

    Entries that say a result needs discussing, a wound needs checking, a fall goes on the falls register or a letter must be marked personal and confidential. They are health information and they decide who gets contacted and how soon. Ahpra says practitioners must apply human judgment to any output of AI. A summary that drops one of these lines, or ranks them differently, can delay care. Patient data guide.

  6. 6
    Labels and opinions about patients

    Words such as frequent flag, third miss and “pushy on the phone” are opinions about identifiable people, and the Privacy Act treats an opinion as personal information whether or not it is true (s 6(1)). The OAIC says inferred or generated information about an identifiable individual that an AI tool produces is also personal information. A chatbot asked to find the patients most likely to miss again creates new information of that kind. Privacy Act guide.

  7. 7
    A voicemail left in error and a possible breach

    Leaving a message that names the practice and the appointment for a patient who asked for no SMS reminders and no voicemail is an unauthorised disclosure if someone else hears it. Section 26WE(2) of the Privacy Act defines an eligible data breach as unauthorised access to or disclosure of information where a reasonable person would conclude it is likely to result in serious harm to any of the individuals concerned. Section 26WH(2) asks an entity that suspects one to carry out a reasonable and expeditious assessment and take all reasonable steps to complete it within 30 days. Whether this is an eligible data breach is for the assessor. A pasted log puts the same incident in a second place. Privacy Act guide.

  8. 8
    The totals row

    Counts of entries and outcomes with no names. The numbers are checkable against the rows (8 did not attend, 3 cancelled late, 1 cancelled with notice) and carry no personal information.

  9. 9
    Staff names and who does what

    Names of the nurse, the practice manager and the receptionist, and the tasks given to them. They are personal information about employees. Under s 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 log into a chatbot is the practice’s act, and may be a disclosure by the practice.

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: Recall letters on Ermos.

Common questions

Do ChatGPT and Claude use the information in a missed appointment log?
Yes, whatever is pasted or uploaded from a log 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 log ties eleven named patients to the appointments they missed, including a woman who could not talk because her partner had arrived, a changed mole result nobody has discussed, and a voicemail left in error after she asked for none.
Can we use ChatGPT or Claude for a missed appointment log if we remove the names?
Not safely, because the reasons and the follow-up notes still point to one person each. 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. A woman of 72 whose husband died on Wednesday, an 81-year-old who fell again on Monday night, or a six-year-old whose mother rang about a rash could be recognised by anyone who knows them. A row linking a person to a health service is health information, which is 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.
Can we upload a missed appointment log spreadsheet to ChatGPT or Claude?
An uploaded log leaves the practice as a whole file. Other tabs such as By clinician and Notes, hidden columns, comments, tracked changes and file properties can travel with the Missed appointments tab. The OAIC's guidance also tells organisations to review carefully the terms and settings that will apply to their use of any AI product. Beyond the rows, the file would carry the mobile numbers, the daughter's call window, the COUNTIF formula and the note that Helen Tran is assessing a possible eligible data breach.
How can a medical practice use AI for a missed appointment log 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. Patterns and a draft missed appointment policy are produced from the practice's own log, and a person checks each letter or message against the flags in it. Helen Tran decides on the breach assessment and a clinician approves the policy.
Is a log of missed appointments health information?
Generally yes. Under section 6FA of the Privacy Act, information about a health service provided or to be provided to an individual is health information, and the same goes for other personal information collected in providing a health service. A row naming a patient and the appointment they missed ties a person to a health service. Health information is sensitive information under section 6(1).
What should a practice do if a message reaches the wrong person or the wrong number?
Treat it as a possible data breach and assess it. Section 26WE(2) of the Privacy Act describes an eligible data breach as unauthorised access to or disclosure of information that a reasonable person would conclude is likely to result in serious harm. Under section 26WH(2) an entity that is aware of reasonable grounds to suspect an eligible data breach must carry out a reasonable and expeditious assessment and take all reasonable steps to finish it within 30 days.
Can a practice put its patients' details into ChatGPT or Claude to analyse no-shows?
The OAIC says that personal information put into an AI system can be a disclosure, which APP 6 restricts, if it is made accessible to others outside the organisation and released from its effective control, and recommends as best practice not entering personal information, particularly sensitive information, into publicly available generative AI tools. NSW health privacy principle 11 also limits disclosure of health information. A practice would need consent or a recognised exception.

See Dominion answering from your own documents.

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

Book a demo