AI and automation for healthcare

Questions and answers

What can AI do for a care organisation or clinic?

Datawonder uses AI in healthcare for the work around care, not for the care itself. Think of a voice agent that books appointments and answers standard questions about preparation and aftercare, an app that lets staff report incidents quickly and an assistant that answers from your own protocols with the source attached. Diagnoses and treatment choices stay with the practitioner.

Can an AI voice agent book appointments for a GP practice or clinic?

Yes. A Datawonder voice agent answers the phone, books, moves or cancels appointments and puts them in the right place in the calendar. It can send reminders and answer standard questions about preparation and aftercare. It does not triage: if the call is about symptoms or medical content, it transfers to a staff member with a summary of what was already said.

How can care staff report aggression incidents more easily?

With a reporting app that keeps reporting short and records the follow-up in one place. For two care organisations Datawonder built a demo that lets staff report aggression incidents and lets the organisation keep track of what happens with each report. There is also the incident reporter: Datawonder's own work that lets care staff report in a few taps and get immediate feedback and guidance.

Can I use AI with patient data?

Yes, but under strict conditions: health data is a special category of personal data under the GDPR. Datawonder records up front which data the system may see, where it is processed, how long it is kept and which data processing agreements are needed. Large-scale processing of health data requires a DPIA. Where possible, patient data stays out altogether.

Which laws apply to a healthcare app or patient portal?

The GDPR always applies. Depending on what the app does, the EU AI Act and the EHDS Regulation for apps that work with health data can apply as well. NIS2 also applies to many healthcare providers. Datawonder uses the applicability scan to work out which of these laws genuinely apply to your app and what has to happen next.

Does the AI make medical decisions?

No. The applications Datawonder builds for healthcare do not make diagnoses and do not give treatment advice: they schedule, report, search and inform. The voice agent does not triage and the assistant on your protocols names the source, so a staff member can check what it says. Software that does help decide on diagnosis or treatment falls under stricter rules, such as those for medical devices.

Can AI answer staff questions about protocols and work instructions?

Yes. Datawonder builds assistants that answer from your own handbooks, protocols and work instructions, with a reference to the source so a staff member can check whether it is correct. The HR assistant on the staff handbook is Datawonder's own work and shows how that works. Before such an assistant answers well, the documents do need to be in order.

How do I know whether my patient portal or healthcare app is secure?

By having it tested the way an attacker would go about it. In a Datawonder penetration test the website or app, the integrations and, if you wish, the source code are examined. You get a report with evidence per finding and criteria for retesting after the fix. Permission is put in writing before anything runs.

How does a care organisation get started with AI?

With one bottleneck you can measure well, such as missed calls or the time reporting takes. Datawonder starts with a free thirty-minute intro call and can then carry out an AI audit: which processes are suitable, what you are better off not automating and what to watch for under the GDPR and the EU AI Act. After that you get a quote with a fixed price, per phase or for the whole project.

Further reading