Please consider supporting us by disabling your content blocker.
loader

FDA Issues First Draft Guidance on AI in Drug Development

The FDA’s New Guidelines on Artificial Intelligence

The FDA on Monday announced its first-ever draft guidance for the use of artificial intelligence (AI) in drug development. This 23-page document marks a significant step in addressing how AI technology should be applied in generating safety, efficacy, and quality data necessary for regulatory decisions.

Central to the FDA’s recommendations is a proposed “risk-based credibility assessment framework” designed to evaluate AI models throughout the drug product life cycle. This framework takes into account various essential factors, such as the question of interest, the context in which the AI model is utilized, risks associated with the model, and the credibility of the AI’s outputs given its specific deployment context.

To illustrate these principles, the FDA provided a hypothetical example. In this scenario, a company is advancing a new drug known to have potential life-threatening side effects. The drug sponsor intends to utilize an AI model to categorize patients into risk-based subgroups based on their likelihood of experiencing these adverse effects. Low-risk patients would receive outpatient monitoring, while those identified as high-risk would be subjected to inpatient surveillance.

According to the FDA, the impact of the model in this situation could be substantial. ‘Model influence would likely be estimated to be high because the AI model will be the sole determinant of which type of patient monitoring a participant undergoes,’ the agency stated. This underscores the potential dangers if a high-risk patient is mistakenly classified for outpatient care. They could face severe adverse reactions without timely intervention in an emergency setting.

Given that AI-generated outputs can significantly affect both regulatory processes and patient outcomes, drug sponsors planning to incorporate AI into their development strategies are urged to provide comprehensive documentation. This includes detailed descriptions regarding the model’s purpose, limitations, and ongoing evaluation.

The FDA also emphasized the importance of early engagement with the agency, encouraging drug developers to “set expectations regarding appropriate credibility assessment activities” from the outset. Collaborative engagement can help identify any potential challenges or obstacles in the intended use of AI during the drug development process.

The FDA is currently accepting public comments on its draft guidance for a period of 90 days and has already begun receiving feedback from industry participants in response to its earlier AI-related discussion paper released mid-2023.

In addition to the AI guidance, the FDA also issued recommendations concerning accelerated approvals and tissue biopsies in clinical trials.