Copper Gluconate/Disulfiram

The FDA is responding to an ever-growing list of medical devices enabled by artificial intelligence and machine learning with new recommendations for updating AI and ML software models with pre-determined change control plans (PCCPs). A key benefit of ML models is that they can learn to work better based on new information over time, but those changes might also introduce safety or efficacy risks in medical devices without sufficient oversight. The FDA wants feedback on the draft guidance by July 3, 2023. The agency published a proposal in 2019 and held public meetings and workshops with the medtech industry, patients and other stakeholders since then. In 2021 the FDA issued an action plan for AI/ML-based software as a medical device (SaMD) products. The new draft guidance promotes the principles of the Blueprint for an AI Bill of Rights released by the Biden Administration last October. The draft guidance is applicable to machine learning-enabled device software functions (ML-DSFs) that the device manufacturer plans to modify over time, whether manually or automatically. “This draft guidance describes an approach that would often be least burdensome and would support the ability to modify an ML-DSF while continuing to provide a reasonable assurance of safety and effectiveness across relevant patient populations,” the FDA said. “Specifically, this draft guidance proposes recommendations on the information to be included in the Predetermined Change Control Plan (PCCP) in a marketing submission for a device that is or includes an ML-DSF.” What you need to know about PCCPs The FDA already has the power to approve or clear PCCPs. The guidance doesn’t apply to PCCPs with only minor modifications that would not require a new submission, but rather device modifications that would otherwise require a premarket approval supplement, de novo submission or a new premarket notification. Manufacturers that include a PCCP in a marketing submission can pre-specify intended modifications and their method of implementation to seek premarket authorization without additional submissions for every change. “In other words, a PCCP, as part of a marketing submission, is intended to provide a means to implement modifications to an ML-DSF that generally would otherwise require additional marketing submissions prior to implementation,” the FDA said in the document. The draft guidance covers the components of a PCCP — description of modifications, modification protocol, and impact assessment — as well as how to establish a PCCP through different regulatory pathways and how to identify a PCCP in a marketing submission. The document also explains how to implement device modifications after a PCCP has been authorized, and how to modify a PCCP for an authorized device. In Appendix B, the FDA offers examples of PCCP scenarios for ML models in software for patient monitoring, skin lesion analysis, ventilator settings, image acquisition assistance and feeding tube placement radiograph analysis. Related: AI breakthroughs in medtech: 7 ways to enhance healthcare