Comment from Sharif Hoque
AnonymousSupportIndividual
Summary: The commenter provides constructive feedback and suggestions to improve the draft guidance on AI in regulatory decision-making. They recommend expanding the scope to include combination products, adding a glossary of terms, and providing more specific examples of model risk categories and GXP impacts.
1. The guidance mentions safety and efficacy for drugs. What about combination products (that involve pharmaceuticals/biologics/medical devices)? These could be mentioned in scope and/or out of scope (Page 1, Line 15)
2. Title of guidance indicates this is for Drugs and Biologics. However the scope only mentions drugs ( Page 3, Line 45)
3. Adding a glossary for terms may be helpful (ie COU, PLR, NPV, etc.)
4. For Questions of Interest, it may be helpful to use list and numbering formatting for each example (Page 6, Line 155)
5. Adding a Model Risk Matrix for different categories may be helpful (Page 8, Line 211).
6. May be helpful to mention how Generative AI may impact this guidance in different ways than traditional AI.
7. May be helpful to mention GXP requirements being impacted by AI.
8. There is an example of High and Medium model risk. It would be helpful to provide a example of low model risk (Page 9, Line 260).
9. Need examples of how Framework for Establishing AI model credibility is impacted by Model Risk category (Page 9, Line 262).
10. Adding a reference to the FDA Digital Health and Artificial Intelligence Glossary would be helpful.
11. May be helpful to mention how corrective and preventative action can be implemented for AI model issues.