Universities
Where research, education and clinical practice meet, we focus on how technology should be evaluated. We consider scientific enquiry alongside the needs of everyday use.
Responsible technology. Better care.
Founded in Tallinn, Estonia in 2021, RAI Health develops and applies artificial intelligence and digital technologies to contribute to better healthcare.
Tallinn, EstoniaThe same purpose, since 2021.
Our story2021Tallinn · Estonia
Building technology that serves healthcare.
RAI Health began in Tallinn in 2021. From the start, our purpose has been to develop technology around the needs of clinicians, healthcare organisations and the people they care for, with responsible AI at the centre of our approach.
We consider imaging, clinical documentation, access to information and digital health workflows together. Our aim is to help the right information reach the right person, support healthcare professionals in applying their expertise, and help organisations work together.
For us, progress is measured by a technology’s place in everyday care as much as a model’s technical capabilities. We put understandable use, meaningful evaluation and clear responsibilities at the heart of development.
Healthcare ecosystem
Different settings. A shared purpose.
RAI Health solutions are used in many universities, hospitals and telehealth companies. Each organisation works differently; they share a need for healthcare technology designed around people.
Hospitals
We look at images, reports and institutional systems as parts of a connected workflow. How clinicians work and how organisations manage their responsibilities guide our design.
Telehealth companies
Remote care depends on access to information, documentation and continuity between teams. We develop technology with the needs of these services in mind.
RESPONSIBLE AI
In healthcare, trust starts with design.
For us, responsible AI means considering who a technology serves, the information it uses and the boundaries within which it operates. We bring patient interests, professional judgement and organisational responsibility into the same conversation.
A fluent or convincing output is not necessarily a correct one. Our development principles therefore look beyond speed and ease of use to the ability to inspect, question and correct a result.
Make room for clinical judgement
We approach AI assistance as a way of working that clinicians can review using their expertise. Visible distinctions between drafts, suggestions and approved reports, the ability to make corrections, and final decisions remaining with the responsible healthcare professional are core design principles.
Consider safety in its intended setting
For every solution, we begin by asking what need it should address and when it should not be used. Results from one setting are not sufficient evidence for another clinical environment. We treat product development and clinical validation as separate activities.
Be clear about limits and uncertainty
Users should be able to understand what information AI has worked from and what it could not access. Making missing context visible, avoiding the presentation of outputs as certainty, and explaining important limitations in plain language are part of our approach.
Treat data with care
When designing a feature, we consider why data is needed, who should access it and how long it should be kept. Images, audio and report text have different needs. We aim to make access, retention and deletion choices understandable.
Evaluate for different people and settings
We consider differences in languages, clinical workflows and patient groups. We favour questioning performance in different conditions over relying on a single average. Translating an interface and validating a clinical model in that language are separate tasks.
Carry responsibility through the product’s life
Responsible AI is not a one-off check. Feedback, error review, assessment of changes and restricting a feature when needed belong in our development approach. Safety and usability need renewed attention as a product evolves.
FROM PRINCIPLES TO PRACTICE
The questions behind every solution.
We assess an idea’s value alongside how it will be used. Four questions guide our development approach.
- 01
What is needed?
What problem are we trying to solve? Who will use the output, which data is necessary, and who holds responsibility?
- 02
What should we evaluate?
Which errors matter? What limits should we examine when information is missing or languages and working environments differ?
- 03
How should it be introduced?
Which review steps, explanations and boundaries are needed? Can users understand what they are approving?
- 04
What needs another look?
What do user feedback and product changes tell us? Is reassessment or a restriction on use needed?
RESPONSIBLE AI IN PRACTICE
What the principles mean in daily work.
How does an AI output differ from an approved report?
AI-generated text may be a draft for a clinician to assess. An approved report follows review by an authorised healthcare professional and the organisation’s reporting process. Our product approach is based on users being able to distinguish these stages.
What is our approach to missing information?
Missing images, limited clinical context or an unclear recording can restrict assessment. Our design aim is to make a gap visible and enable the necessary review. For example, assessing selected images should not be presented as reviewing a complete study.
Does announcing a feature mean it is ready for clinical use?
Development, pilot testing and validation for a particular use are different stages. Availability, clinical scope and the necessary assessments depend on the product and its setting. This page describes our company approach, not a blanket clinical authorisation or certification for every product.
Why is responsible AI about more than choosing a model?
A capable model alone does not create a good healthcare workflow. Data selection, access permissions, user training, presentation of outputs and the handling of errors all matter. We therefore consider AI together with the interface, the process and human responsibility.
Let’s develop the future of care responsibly.
Tell us about the needs of your university, hospital or telehealth service. Explore our solutions and the products we are developing.