Page 27 - Shaping ethics, regulation and standardization in AI for health

Shaping ethics, regulation and standardization in AI for health

A�3 Technical

A�3�1 DEL 0�1: Common unified terms in artificial intelligence for health

Summary: This deliverable contains a glossary with agreed terminology in artificial intelligence
(AI) for health for use not only across the various FG-AI4H Deliverables, but also to promote
the harmonized use of important AI for health terms across the different disciplines involved in
this cross-disciplinary field.

A�3�2 DEL 3: AI4H requirement specifications

Summary: The purpose of this document is to define the lifecycle-based System Requirement
Specifications (SyRS) that explains the informational, functional, behavioural and operational
aspects of a generic AI for health (AI4H) system.

SyRS serves as the basis for the system design, system verification and validation plans and
procedures for the AI4H system.

System requirements analysis methodology follows a collaborative team-oriented approach,
involving all the working groups and topic groups of FG-AI4H, to help the project team identify,
control and track various requirements and changes to those requirements during the AI4H
system development lifecycle.

Tables are intended to serve as checklists for configuring a basic minimal set of ML4H system
/ product lifecycle requirements specifications, which include the technical, the clinical, the
regulatory and the ethical requirements. In the ML4H system / product testing phase, the
same tables can be used to generate applicable test cases for verification of requirements
specifications to support ML4H product conformity assessment procedures.

A�3�3 DEL 4: AI software life cycle specification

Summary: Building a quality product includes performing quality tasks throughout the
development lifecycle. For example, having a plan that describes the process that you use to
develop software is better than not having a plan. Having product requirements in a documented
form is better than having product requirements only in people's minds. Documenting the design
of the software at both a high-level architecture as well as a unit-level helps tremendously when
trying to maintain a piece of software that was written years ago, and the original developers
are no longer available to help.

Due to the "black box" nature of some machine learning (ML) applications, having quality
processes in place will be a significant factor affecting product quality and performance.

DEL4 provides an overview of existing software development lifecycle standards and how they
can be applied to the development of health artificial intelligence (AI) applications.

A�3�4 DEL 5�1: Data requirements

Summary: Deliverable 5.1 lists acceptance criteria for data submitted to the FG-AI4H
benchmarking platform and states the governing principles and rules. These principles are
crucial because the core of the benchmarking framework for AI for health methods will be an

17