5th Global Gastroenterology and Artificial Intelligence Summit (Live)

6 - Scheiman

Pdf Summary

This document discusses the topic of AI-enabled devices and their regulation by the FDA. The author emphasizes that patients should always be at the heart of device development, ensuring access to high-quality, safe, and effective medical devices. The FDA uses a risk-based approach when evaluating AI products, considering factors such as intended use, benefits, and risks. Developers of AI devices should assess and mitigate risks, demonstrate the benefits of their devices through performance testing and clinical studies, and ensure transparency in labeling. <br /><br />The document also provides an overview of the different classes of medical devices based on their risk level. Class I devices have general controls that ensure safety and effectiveness but are not life-sustaining or of substantial importance. Class II devices require both general and special controls, such as performance standards and post-market surveillance. Class III devices, which are high-risk, require general controls and premarket approval. <br /><br />To market their products, developers can go through the 510(k) pathway, which demonstrates substantial equivalence to a legally marketed device, or the PMA pathway, which requires more stringent data requirements for higher-risk devices. The author explains that a 510(k) submission is required when introducing a device for the first time, changing its indications for use, or making modifications that could affect safety or effectiveness. PMA reviews require valid scientific evidence to assess safety and effectiveness. <br /><br />The author also emphasizes the importance of considering factors such as benefits, risks, patient perspective, and post-market data when making benefit/risk determinations. Additionally, the FDA considers the type, magnitude, probability, and duration of benefits, as well as the severity, types, and rates of harmful events when assessing risk. Other factors taken into account include uncertainty, availability of alternative treatments, and risk mitigation. <br /><br />In summary, the document highlights the increasing impact of AI/ML-enabled devices in healthcare and the importance of collaboration between developers and the FDA. It provides resources for digital health technologies and emphasizes the need for favorable benefit-risk assessments for device authorization.

Keywords

AI-enabled devices

FDA regulation

patient-centric development

device safety

device effectiveness

risk-based approach

medical device classification

510(k) pathway

PMA pathway

benefit-risk determinations