AI in the HOF Clinic: Uses and Limits
Artificial intelligence (AI) tools have entered the routine of many clinics: they write caption drafts, summarize texts, organize schedules, answer simple questions and generate images. For the orofacial harmonization (HOF) professional, the useful question is not "will AI replace me?" but "which tasks does it save time on without adding risk, and where should I not even touch it?".
This guide separates uses that usually make sense, uses that call for care and uses AI should not take on, with simple criteria to decide. It does not recommend any specific product and does not replace legal guidance on data protection and professional council rules (in Brazil, the LGPD and councils such as CFO and CFM).
Where AI usually helps
Repetitive text and organization tasks, where a mistake is easy to spot and fix:
| Use | How it helps | Care needed |
|---|---|---|
| Caption and script drafts | Gives you an initial structure to edit | Check against council rules. See ethical captions for HOF |
| Summaries of long texts | Speeds up reading articles and regulations | Check the source. AI can be wrong or make things up |
| Editorial calendar | Suggests topics and dates | Adapt it to your reality. See editorial calendar |
| Message drafts | First version of reminders and replies | Personalize and remove promises |
| Spreadsheet organization | Formulas, templates, data cleanup | Don't include identifiable data unless necessary |
| Ideas for FAQs | Lists questions patients tend to have | Answer them yourself, on a clinical basis |
In all these cases the rule is the same: AI produces a draft, and responsibility for what goes out belongs to the professional. A text it produces with apparent confidence can contain errors, unsupported claims or promises your rules prohibit.
Where it calls for extra care
Patient data and photos
Health data is sensitive. If you paste a name, concern, patient record or photo into an AI tool, you are sending personal data to a third party. That requires a legal basis, a clear purpose and, generally, specific consent. See LGPD in the clinic's patient records and photos and consent for AI with patient photos.
A practical rule: before using the tool, ask yourself whether it would be acceptable for the patient to read exactly what you sent. If not, anonymize it or don't send it.
Automated service
Chatbots and automatic replies help with schedule hours, address and simple questions. But clinical triage, symptom guidance or answers about a complication cannot be delegated to a bot. If the patient reports pain, a change in skin color, a visual change or any sign of a complication after a procedure, the path is immediate human contact. See chatbot in clinic service.
Generated images
An AI-made image (of a fictional person or of a simulation) must be identified as such, and can never be presented as a real patient's result. This applies to your profile, your website and your printed material. See what you can advertise.
Where AI does not replace the professional
Some decisions should not be handed to AI, however convincing the answer looks:
- Indicating or contraindicating a procedure.
- Defining volume, product or technique of application.
- Assessing anatomy, vascular risk and the patient's conditions.
- Managing a complication: occlusion, infection, allergic reaction.
- Obtaining informed consent, which is a human act of explaining and listening. See informed consent.
- Judging whether the patient has realistic expectations or signs of dissatisfaction with their own body beyond what is expected.
AI can help organize your reasoning, but it doesn't examine the patient, doesn't know the full history and answers for nothing. You are the one who answers.
Where visual simulation fits on this map
AI visual simulation is a case of its own, because it combines sensitive data (the face) with the patient's expectations. Used with consent and honest language, it helps the patient understand the plan. Used without notice or as a promise, it becomes a problem. The basic rules:
- Patient notice and acceptance before sending the photo.
- A clear simulation label on every shared image.
- A conversation about limits: the image is not a guarantee. See limits of visual simulation in the consultation.
- No public advertising without specific authorization.
Internal policy: simple and in writing
If the clinic has a team, a short AI policy is worthwhile. It can fit on one page and answer:
- Which tools are allowed.
- What data can never be sent (name, photo, patient record, phone number).
- Who reviews what the AI produces before it is published.
- How to record the patient's acceptance when the tool receives an image.
- Whom to turn to in case of doubt or an incident.
This protects the patient, the team and the clinic. See the HOF clinic process manual to find where to fit this policy and the team social media policy template.
Questions to test your judgment
Common mistakes
- Publishing AI text without reading it.
- Sending a name, photo or patient record without telling the patient.
- Treating the simulation as a result.
- Delegating complication triage to a bot.
- Using AI to define volume or treatment.
- Not knowing where the data goes.
- Having a team using different tools, with no rule.
How Harmy fits in
Harmy uses AI at one specific point: generating, from a front photo and the ml marked per area, an illustrative projection to support the conversation in the consultation. The professional gives their acceptance in the app before the photo is sent to a third-party AI, the face contours are calculated on the device and are not stored, and the image comes out with a simulation notice. The decision about what to treat, how much and how remains the professional's.
Summary
Use AI to save time on drafts and organization, with review. Protect patient data, label simulation as simulation and keep everything clinical, ethical and consent-related with you. The tool serves the consultation; it doesn't decide for it.
Frequently asked questions
Can artificial intelligence replace the HOF professional?
No. Assessment, indication, performing the procedure, managing complications and clinical decisions are the responsibility of the licensed professional. AI is a support tool for text, organization and communication tasks.
Can I use AI to write captions and messages?
You can, as long as you review everything. The final text is your responsibility and must follow your professional council's advertising rules. Never publish without reading, and don't let the AI invent data or promises.
Is it safe to put patient data into an AI tool?
Only with care. Health data requires a legal basis, consent and knowing how the tool handles the information. When in doubt, don't send a name, photo or patient record; use anonymized data.
Is an AI simulation the result of the procedure?
No. It is an illustration to support the conversation. The actual result depends on individual factors and may be different. Always make this clear to the patient.
Do I need to tell the patient I use AI?
When the tool receives the patient's data or image, yes, clearly and specifically, with a record of the acceptance. For internal use with no personal data, the notice doesn't apply in the same way.