Steps to release an AI/ML based medical device in FDA regulated environment!

May the 4th was with me today. So I found the force to write this article today :). Note that FDA routinely updates their guidance documents. Please let me know if you feel any corrections are needed. Be sure to review the latest (March 2024) paper from FDA https://www.fda.gov/media/177030/download?

Here's a general outline of the process:

1.????? Preliminary Research: Before beginning development, conduct thorough research to ensure that your device addresses a genuine medical need and that there's a market for it.

2.????? Patents and Intellectual Property (IP): It is always a good idea to perform say a Google Patent search to find out if your research is unique and original enough to file for a patent. Originality could be in terms of the AI/ML algorithm itself, the data processing technique or the application. Consult your IP counsel when in doubt.

3.????? FDA Regulations Familiarization: Become familiar with FDA regulations relevant to medical devices, particularly those pertaining to AI/ML applications. Here is an example:

• FDA's Guidance for Industry and FDA Staff: "Clinical Decision Support Software" (2019) https://www.fda.gov/media/109618/download

4.????? Quality Management System (QMS) Implementation: Establish a QMS compliant with FDA regulations, such as ISO 13485, to ensure that your device is developed, manufactured, and distributed in a controlled manner.

• https://www.federalregister.gov/documents/2024/02/02/2024-01709/medical-devices-quality-system-regulation-amendments

5.????? Risk Management: Conduct risk assessments throughout the development process to identify and mitigate potential hazards associated with your device.

• ISO 14971: Medical devices — Application of risk management to medical devices https://www.iso.org/standard/72704.html

6.????? Data Collection and Validation: Gather data to train and validate your AI algorithms. Ensure that the data is representative, high-quality, and appropriately labeled.

7.????? Algorithm Development and Validation: Develop and validate your AI algorithms according to best practices, ensuring robustness, accuracy, and generalizability.

• See FDA’s Good Machine Learning Practices: https://www.fda.gov/medical-devices/software-medical-device-samd/good-machine-learning-practice-medical-device-development-guiding-principles

8.????? Software Development Life Cycle (SDLC) Compliance: Develop your software following an SDLC process that complies with FDA regulations, including requirements for software verification and validation.

• FDA's General Principles of Software Validation; Final Guidance for Industry and FDA Staff (2002) https://www.fda.gov/media/73141/download
• Global Approach to Software as a Medical Device https://www.fda.gov/medical-devices/software-medical-device-samd/global-approach-software-medical-device

9.????? Clinical Validation: Conduct clinical studies to demonstrate the safety and efficacy of your AI/ML-based medical device. Ensure that the study design meets FDA requirements for clinical evidence. Here are some examples:

• FDA's Guidance for Industry and FDA Staff: "Acceptance of Medical Device Clinical Data from Studies Conducted Outside the United States" (2019) https://www.fda.gov/media/111346/download
• FDA's Guidance for Industry and FDA Staff: "Investigator Responsibilities — Protecting the Rights, Safety, and Welfare of Study Subjects" (2009) https://www.fda.gov/media/77765/download

10.?? Regulatory Submission: Prepare and submit a regulatory application to the FDA. The type of submission (e.g., 510(k), De Novo, PMA) will depend on the classification of your device and the level of risk it poses.

• FDA's Guidance for 510K: https://www.fda.gov/medical-devices/device-approvals-and-clearances/510k-clearances
• FDA's Guidance for De Novo: https://www.fda.gov/medical-devices/premarket-submissions-selecting-and-preparing-correct-submission/de-novo-classification-request
• FDA's Guidance for PMA: https://www.fda.gov/medical-devices/premarket-submissions-selecting-and-preparing-correct-submission/premarket-approval-pma

11.?? FDA Review: The FDA will review your submission, which may involve requests for additional information or clarification. Be prepared to respond promptly to any FDA inquiries.

• FDA's Guidance for Industry and FDA Staff: "Medical Devices: The Pre-Submission Program and Meetings with FDA Staff" (2014) https://www.fda.gov/media/93740/download

12.?? Post-Market Surveillance: Implement a post-market surveillance plan to monitor the safety and performance of your device once it's on the market. Report any adverse events to the FDA as required. See examples for guidance on changes to medical devices:

• FDA's Guidance for Industry and FDA Staff: "Changes to Existing Medical Software Policies Resulting from Section 3060 of the 21st Century Cures Act" (2019) https://www.fda.gov/media/109622/download
• Predetermined Change Control Plans (PCCP) for Machine Learning Enabled Medical Devices https://www.fda.gov/media/173206/download

13.?? Labeling and Marketing: Ensure that your device labeling and promotional materials comply with FDA regulations. Only make claims that are supported by scientific evidence.

14.?? Distribution and Post-Market Compliance: Establish procedures for distribution and post-market compliance, including complaint handling, device tracking, and field corrective actions.

15.?? Continual Improvement: Continuously monitor and improve your device based on feedback from users, clinical experience, and post-market surveillance data.