FDA establishes guidelines for AI openness in medical devices

In 2021, the U.K.’s Medicines and Healthcare Products Regulatory Agency, Health Canada, and the FDA collaborated to develop ten guiding principles for best practices in artificial intelligence. However, they did not provide a thorough analysis of how to meet that need.

The three agencies have now released a second paper that focuses on openness, something they describe as the extent to which pertinent audiences are given straightforward details about a medical device that uses machine learning.

Programmes can produce an output without stating how they arrived at their result, for example a diagnostic advice. If a gadget can explain the reasoning behind its results, it is said to have “explainability”.

The FDA stated that “when this information is available and easily understood,” it is desirable to allow people access to the algorithm’s rationale. When making decisions about patient care, users can utilise the data to “critically analyse the device and its output.”

Device developers are asked to take into account the who, why, what, where, when, and how of openness by the FDA and its partners.