The Food and Drug Administration (FDA) recently issued new draft guidance for developers of medical devices that incorporate artificial intelligence (AI). The guidance focuses on the need for comprehensive information regarding the data used to train and validate these devices.
Specifically, the FDA is requesting more detail on data sources and the demographics of the data sets. The agency also wants developers to disclose potential blind spots and biases that could impact device performance. This information is to be included in marketing submissions to the FDA.
This new guidance follows the authorization of over 1,000 AI devices by the FDA. The guidance does not mandate additional testing in live clinical settings as a condition for approval. However, it does recommend that manufacturers submit a plan for monitoring their devices’ performance in clinical settings after commercial clearance.