Machine Learning
Product Development

FDA, MHRA, Health Canada AI Update: Transparency Princples for Machine Learning-Enabled Medical Devices

Transparency to enhance safety and effectivenes of ML-enabled medical devices

cosm logo
guiding principles for transparency of machine learning-enabled medical devices

Transparency Principles for Machine Learning-Enabled Medical Devices


In June 2024, the U.S. Food and Drug Administration (FDA), Health Canada, and the UK's MHRA released guiding principles to enhance transparency for machine learning-enabled medical devices (MLMDs). These principles build upon the FDA's 2021 Good Machine Learning Practice (GMLP) principles, particularly those emphasizing the performance of the human-AI team (principle 7) and the provision of clear, essential information to users (principle 9).

Definition of Transparency

Transparency for MLMDs involves the clear communication of appropriate information about the device, including its intended use, development, performance, and, when available, the logic behind its operations. This encompasses both the concept of “logic” (how an output or decision was reached) and “explainability” (the degree to which this logic can be understood by a person).

Key Aspects of Effective Transparency

Effective transparency for MLMDs ensures that information impacting risks and patient outcomes is communicated, considers the needs and contexts of intended users, utilizes the best media and strategies for communication, and relies on a holistic understanding of users, environments, and workflows.

Human-Centered Design

The guiding principles outline that transparency is is based on the critical concept of “human-centered design,” which involves an iterative process addressing the entire user experience and involving relevant parties throughout design and development. This approach helps develop MLMDs with high transparency, validates transparency, and ensures that users have all necessary device-related information.

Guiding Principles

The guiding principles for transparency of MLMDs focus on the following areas:

1. Who: Relevant Audiences

Transparency should cater to:
  • Users of the device, such as healthcare professionals, patients, and caregivers.
  • Those receiving healthcare with the device, including patients.
  • Additional parties like support staff, administrators, payors, and governing bodies

2. Why: Motivation

Transparency is essential for:
  • Ensuring patient-centered care.
  • Supporting the safe and effective use of the device.
  • Identifying and evaluating device risks and benefits.
  • Promoting health equity by identifying and assessing bias.
  • Supporting the maintenance and continued safety of the device.
  • Fostering trust and confidence in the technology, thereby encouraging adoption and access.

3. What: Relevant Information

The information shared should enhance understanding of the device and its intended use. This includes:
  • Medical purpose and function of the device.
  • Diseases or conditions it addresses.
  • Intended users, use environments, and target populations.
  • How the device fits into the healthcare workflow.
  • Details about device performance, benefits, and risks.
  • Risk management activities, including bias management strategies.
  • The logic behind the device’s outputs, when understandable.
  • Information on product development, clinical studies, and lifecycle risk management.

4. Where: Placement of Information

Device information should be easily accessible through the user interface, which includes all elements the user interacts with (e.g., training materials, controls, displays, packaging, labeling, and alarms). The software user interface should be optimized to make information responsive, personalized, adaptive, and reciprocal.

5. When: Timing of Communication

Effective transparency involves considering information needs throughout the product lifecycle, including:
  • Providing detailed information when acquiring or implementing the device.
  • Timely notifications of device updates or new information.
  • Targeted information during specific workflow stages or triggers.

6. How: Methods to Support Transparency

Supporting transparency involves a holistic understanding of users, environments, and workflows. This can be achieved through human-centered design principles, which ensure information is accessible and usable by:
  • Providing appropriate detail for the intended audience.
  • Arranging content to support user decision-making.
  • Using plain language or technical language as relevant.

Checkout the transparency principles from the FDA here:


The guiding principles for transparency of MLMDs are designed to ensure that appropriate information is clearly communicated to relevant audiences. This helps in making informed decisions, maintaining device safety, fostering trust in technology, and supporting the adoption of beneficial innovations. Continuous engagement and feedback are encouraged to evolve and refine these principles in the rapidly advancing field of machine learning-enabled medical devices.

Image Source: Created with assistance from ChatGPT, powered by OpenAI

Disclaimer -