Hospital AI Lab Practical information on medical AI for Irish hospitals

Updated:

Practical tips

These are the author’s practical suggestions, not legal requirements. Where a document applies, the page says so and names it.

Before using any of the tools listed here for work, make sure your hospital’s information security policy allows it. Do not enter patient or staff data into outside services: that is the general rule for every tool.

Do not paste patient information into tools you have not approved

Names, PPS numbers, dates of birth, phone numbers and the contents of records should stay out of consumer chatbots. The Data Protection Commission’s AI guidance warns that personal data typed into AI tools can raise security and data protection risks, and that some models can reproduce training data. Say clearly which tools are approved.

Keep a list of AI tools and a name beside each

HIQA’s January 2026 draft suggests a list of authorised AI tools shared with staff and an individual assigned to keep an inventory. A shared spreadsheet is a fine start.

Check AI drafts the way you would check a trainee’s first draft

AI can state wrong things confidently. The Data Protection Commission notes outputs relied on without human review create automated decision-making risks. Check facts, numbers and names against the source before anything enters a record.

Ask vendors three questions in writing

Is it a medical device and why or why not? Where does our data go and is it re-used? How will you tell us about a fault or a software update? Keep the answers with the contract.

Put the clocks on one card

A serious device incident goes to HPRA (HPRA sets no hospital deadline, so decide your own). A personal data breach has a 72-hour clock for the controller under GDPR Article 33. A serious incident with a high-risk AI system goes to the provider first, then to the authorities (AI Act Article 26(5)). Put who does what on one page.

Tell patients when AI is part of their care

HIQA’s draft and the AI for Care strategy both stress transparency about when, where and why AI is used. Decide in advance who explains it and what the short answer is.