FDA Continues the Discussion on AI/ML

Many of our organizations are somewhere in the journey of using AI/ML some where in the drug product lifecycle, so it is no surprise that the FDA is continuing the dialogue with the recently published draft of “Considerations for the Use of Artificial Intelligence to Support Regulatory Decision-Making for Drug and Biological Products.”

This draft guidance lays out a solid approach by using a risk-based credibility assessment framework to establish and evaluate the credibility of AI models. This involves:

• Determining if the model is adequate for the intended use
• Defining the question of interest the AI model will address
• Defining the context of use for the AI model
• Assessing the AI model risk based on model influence and decision consequence
• Developing a plan to establish model credibility commensurate with the risk
• Executing the plan and documenting results

I think may of us are in the midst of figuring out how to provide sufficient transparency around AI model development, evaluation, and outputs to support regulatory decision-making and what will be found to be acceptable. This sort of guidance is a good way for the agency to further that discussion and I definitely plan on commenting on this one.