TRUSTWORTHY AI FOR PSYCHIATRY

Trust starts with
clear answers.

You need to know how source material is handled, what the tool does and what you are responsible for. That is the starting point for our documentation support for psychiatry.

01 / HUMAN OVERSIGHT

You review.
You approve.

Summarizer compiles documentation drafts from your source material. You check information against the sources and decide what to take forward. Clinical judgement remains yours.

See the review process in practice →
02 / OUR OWN AI INFRASTRUCTURE

Our own hardware
in Sweden.

Our AI processing runs on our own hardware in Norrbotten. This gives us control over that part of the process. Infrastructure and subprocessors for other parts of the service are described in the customer documentation.

Watch the video about our infrastructure →
03 / DATA AND RESPONSIBILITIES

Clear terms
for your source material.

Your organisation needs to establish permitted use, access permissions and agreements. Working material may be stored temporarily and is subject to retention and deletion rules. Summarizer is not a patient archive.

Read about data and agreements →
04 / A CONSIDERED WAY OF WORKING

Understand the tool
before you start.

The introduction covers both the workflow and its limitations. Practise checking sources, spotting possible errors and understanding your responsibilities through videos, text and short questions.

Take the introduction →

Where does your source material go?

About our own hardware and AI processing in Norrbotten. 1 min 11 sec · Swedish narration and on-screen text. English summary below.

The video's key messages in text
  • Summarizer creates documentation drafts from your material. You review, edit and approve them.
  • Careifai's AI processing runs on our own hardware in Sweden, based in Norrbotten.
  • Operating our own infrastructure gives us control over AI processing and reduces dependence on external AI platforms.
  • Deployment within the healthcare provider's own environment is described as a future option.
Open the video separately
European Union emblemEUROPEAN UNION REGULATION

The EU
AI Act.

The EU's framework for trustworthy AI, centred on people's safety and rights.

Read about the AI Act in a new tab

OUR TEAM'S EXPERTISE

Zahoor Ul Islam, CTO.

Zahoor contributed survey analysis to AI4EU's methodology for value-based AI design. The contribution appears in the version history of the report AI4EU Value-based Design Methodology, published in 2021 within the EU research programme.

Experience of incorporating human values into AI design is relevant to responsible AI questions also addressed by the EU AI Act.

AI4EU was a research project. Participation is not product approval or certification under the EU AI Act.

Questions before you start.

What does your own AI infrastructure mean for personal data?

AI processing on our own infrastructure in Sweden is one part of how the service operates. It does not replace the need for a legal basis, access permissions, agreements and oversight of the entire processing chain. The customer documentation also describes other infrastructure, subprocessors, temporary storage and deletion. Ask us for current documentation to inform your assessment.

Can we start using patient material straight away?

Your organisation first needs to clarify decisions, procedures and the applicable data processing agreement. An account or a completed introduction does not replace these steps. Start with fictional practice material. De-identification does not guarantee that information has become anonymous.

How do we get security and contractual documentation?

Contact info@careifai.com and describe your organisation and the use you are considering. Ask for current customer documentation, a data processing agreement and details of infrastructure and subprocessors. Do not include patient information in your enquiry.

A CONTEXT FOR DEVELOPMENT

Networks, collaboration and support.

See how it fits your work.

Five short steps on source material, drafts and review.

Learn to use Summarizer