Transparency Guidelines for Medical Devices with Artificial Intelligence

After the FDA, together with Health Canada and the MHRA from the UK, published 10 guidelines on Good Machine Learning Practice (GMLP) two years ago (also see Dealing with artificial intelligence in the context of the further development of software in medical devices), transparency guidelines for medicinal products with AI are now following.

In this document, the term "transparency" describes the extent to which appropriate information about machine learning-enabled medical devices (MLMDs) is clearly communicated to those affected.

Transparency ensures

• that information that could impact risks and patient outcomes is communicated
• takes into account the information needed by the intended user or audience and the context in which it will be used 
• uses media, timing and strategies for successful communication
• is based on a holistic understanding of users, environments and workflows.

Another important concept related to transparency is "human centered design". The FDA understands this to be an iterative process that focuses on the user experience and involves the relevant parties throughout the design phase of development. The goal is:

• to develop MLMDs with a high degree of transparency
• Assist in the validation of transparency
• Ensure that users receive all the device-related information they need.

The principles of the guideline

The principles of the guidelines are based on following questions:

• Who (relevant target groups)
• Why (motivation)
• What (relevant information)
• Where (placement of the information)
• When (timing)
• How (methods to promote transparency)

And it is precisely these 6 questions that are explained in individual sections of the guideline. Despite these explanations, the content of these explanations is then summarized again in a table.

You can find the transparency recommendations on the CDRH website.