Page 18 - Shaping ethics, regulation and standardization in AI for health

Shaping ethics, regulation and standardization in AI for health

– DEL 10.24: FG-AI4H Topic Description Document for the Topic group on AI for point-of
care diagnostics (TG-POC)
– TG-Dental Output 1: Artificial intelligence in dental research: A checklist for authors and
reviewers
– TG-Dental Output 2: Artificial intelligence for oral and dental healthcare: Core education
curriculum
– TG-Dental Output 3: Ethical considerations on artificial intelligence in dentistry:
A framework and checklist
– AHG-DT4HE Output 1: Guidance on digital technologies for COVID health emergency

4 Open Code Initiative (OCI)

The Open Code Initiative (OCI) of the FG-AI4H developed a running platform with the proof-
of-concept for the assessment and benchmarking of AI for health solutions using the concepts
identified in the various FG-AI4H Deliverables. The OCI developed the digital building blocks
of the FG-AI4H assessment platform, which can support the end-to-end development and
assessment of AI for health algorithms under consideration of regulatory guidelines and the
needs of all AI for health stakeholders.

The platform is intended to help in developing and assessing AI4H products, and to provide
guidance to implementers developing their own applications. It can also be used by other
stakeholders, such as regulatory bodies and medical professionals. OCI's open code nature
aims to allow reuse in a variety of contexts, as well as for transparency and easier verification.

The OCI was structured around six working packages (Figure 3 illustrates the evaluation
package):

– Core package: Provisions the common services to all packages, e.g., storage,
authentication and authorization to access resources.
– Data acquisition and storage package: Provides safe and secure storage of medical data;
serves as an interface for other packages to access this data using the FHIR standard;
and offers data governance that complies with data protection laws. Facilitates data
compilation for many modalities; registers data and metadata; offers a federated data
catalogue and data governance; ingests data; and manages patient consent information.
– Data annotation package: Provides an annotation campaign management API and
annotation interfaces for many modalities; includes collaboration features; develops a
network of annotation experts; and creates notifications of pending annotation tasks. OCI
partners provide annotation tools integrated with the FG-AI4H platform, e.g., AI-based
pre-annotation platform or 3D image annotation user interface.
– Training package: Provides a Privacy Preserving Federated Learning solution enabling
training on datasets that need to stay at their location. This solution guarantees secure
sharing of models and model aggregation.

8