Artificial Intelligence Algorithms in Health Care: Is the Current Food and Drug Administration Regulation Sufficient?

Area of Science:

• Healthcare Regulation
• Medical Artificial Intelligence
• Machine Learning in Medicine

Background:

• Machine learning (ML) is increasingly used in healthcare, posing regulatory challenges.
• Current Food and Drug Administration (FDA) regulations for ML algorithms hinder adaptation to clinical environments.
• Ensuring patient safety is paramount when ML algorithms support or replace medical professionals.

Purpose of the Study:

• To propose a novel regulatory pathway for ML algorithms in healthcare.
• To address the limitations of current regulatory frameworks for adaptive ML technologies.
• To ensure the safe and effective integration of ML into clinical practice.

Main Methods:

• Literature review of AI, ML, and regulation in healthcare from 2017-2022.
• Proposal of a new regulatory framework analogous to medical professional regulation.
• Discussion of technical and non-technical implementation challenges and solutions.

Main Results:

• Current regulations prevent ML algorithms from adapting to real-world clinical feedback.
• A proposed lifecycle-based regulatory pathway offers a solution for adaptive ML.
• Identified challenges and potential solutions for implementing the proposed pathway.

Conclusions:

• A novel regulatory approach is needed to govern the lifecycle of ML algorithms in healthcare.
• This approach can foster innovation while maintaining patient safety, quality, and equity.
• Addressing implementation challenges is crucial for realizing the full potential of ML in healthcare.