EU AI Act & Responsible AI
1. Purpose
eDoer uses AI-powered functionality to support learning and, where enabled, Patient Education.
eDoer aims to use AI in a transparent and responsible manner and to implement applicable requirements of the EU Artificial Intelligence Act (EU AI Act).
This document describes the initial measures implemented by eDoer, with a particular focus on AI transparency, AI literacy and responsible use of AI in Patient Education.
2. AI Transparency
Clear Disclosure of AI Interaction
eDoer clearly informs Users when they are interacting directly with AI-powered functionality.
For Patient Education, the AI interface is clearly identified as AI-powered, and Users are informed from the beginning of the interaction that they are communicating with an AI system.
For example:
AI Patient Education Assistant
You are interacting with an AI-powered assistant.
Where appropriate, additional information may be provided to clarify the role and limitations of the AI Assistant:
This assistant supports patient information and learning and does not replace professional medical advice, diagnosis or treatment.
The disclosure is presented directly within the User experience and is not limited to Terms and Conditions, Privacy Policies or other legal documentation.
This approach supports the transparency requirements for AI systems interacting directly with natural persons under Article 50 of the EU AI Act.
Implementation Evidence
eDoer maintains evidence that the AI disclosure is implemented within the User interface, for example through screenshots or records of the relevant interface implementation.
3. AI Literacy
eDoer takes measures to support AI literacy among its personnel who are involved in developing, configuring, operating or supporting AI-powered functionality.
Relevant eDoer personnel are informed about topics appropriate to their role and interaction with AI, including:
- the purpose and capabilities of the AI functionality used by eDoer;
- the limitations of AI and the possibility of inaccurate, incomplete or unsupported AI-generated information;
- appropriate handling of personal and health information when using AI functionality;
- the defined boundaries of Patient Education AI; and
- the importance of responsible use of AI in a healthcare context.
Personnel working directly with Patient Education AI are expected to understand these principles and the intended role and limitations of the functionality.
eDoer communicates these principles to relevant personnel and maintains appropriate internal guidance or records of such initiatives.
These measures support the AI literacy requirements under Article 4 of the EU AI Act.
4. Responsible AI Boundaries for Patient Education
In addition to the specific transparency and AI literacy measures described above, eDoer applies defined functional boundaries to support the responsible use of AI in Patient Education.
eDoer Patient Education AI is designed to support activities such as:
- accessing and retrieving relevant healthcare information;
- explaining healthcare information and terminology;
- summarizing information and documents;
- translating information;
- supporting patient learning and understanding; and
- answering educational questions based on permitted information sources.
The Patient Education AI is not intended to:
- diagnose medical conditions;
- select or recommend treatments;
- recommend starting, stopping or changing medication;
- calculate patient-specific clinical risk;
- monitor physiological parameters; or
- make clinical decisions on behalf of healthcare professionals.
Healthcare professionals remain responsible for medical diagnosis, treatment and clinical decision-making.
eDoer applies these boundaries when designing, configuring and testing its Patient Education AI functionality.
5. Implementation & Evidence
eDoer maintains proportionate evidence supporting the measures described in this document.
For the current implementation, this includes:
AI Transparency
Evidence that Users are clearly informed when they interact directly with AI-powered functionality, such as screenshots or records of the relevant User interface.
AI Literacy
Internal guidance and communication of relevant AI capabilities, limitations, risks and responsible-use principles to eDoer personnel involved with AI functionality.
Responsible AI Boundaries
Defined functional boundaries for Patient Education AI, reflected in the design, configuration and testing of the functionality.
These measures may be reviewed and updated as eDoer's AI functionality and applicable regulatory requirements evolve.
