Can I use AI for Quality and Regulatory?

The use of artificial intelligence (AI) is a hot topic in the medical device industry today.  With the financial and time pressures that many medical device companies are under, it can be tempting to use artificial intelligence to replace the expertise provided by consultants. While AI can be a useful tool to obtain and summarize information as well as generate documentation, overreliance on it can be dangerous. Knowing when and how to use AI for quality and regulatory purposes can help you utilize it effectively.

The Problem with AI

1.      AI only knows what you tell it

A quality management system is a complex set of quality manuals, standard operating procedures, work instructions, forms, and records. But, questions asked to AI typically only submit a small portion of information. This can result in AI generating content that already exists in other places in your QMS, content that conflicts with information in other places in your QMS, or content that is not correct.

2.      AI doesn’t always know the latest in medical device standards and regulations

Regulation and standards requirements for medical devices are constantly changing. One of the benefits of AI is that it mines information from multiple sources and mitigates the risk of relying on a single unreliable source. Unfortunately, this means that AI mines both outdated and current requirements and can generate responses that don’t comply with current expectations. This is especially true for recently revised requirements since the AI model training material on the old requirements outweighs the AI model training material on the new requirements.

3.      Public AI clients can mine confidential information

If you use public large language models, they can mine any information you submit to it and train their models using it. This can result in your company’s confidential information being shared in AI results provided to other companies.

When Should I Use AI?

1.      To identify primary sources 

Not sure which standards or regulations govern biocompatibility testing? Ask AI to identify those sources for you, then read those standards and regulations yourself to understand the requirements.

2.      To generate a first draft

If you’re using a private instance of an AI model, submitting relevant information to an AI model to generate a first draft of a document can be helpful. However, a human will still need to review the AI-generated draft to ensure it meets applicable requirements, communicates information succinctly in a way that is understandable to regulatory agencies, and works with existing documentation in your quality management system. A recent warning letter issued by the FDA to Purolea Cosmetics Lab states that “if you use AI as an aid in document creation, you must review the AI generated documents to ensure they were accurate and actually comply” with applicable requirements. In this warning letter, the FDA makes it clear that relying only upon AI generated content is not permitted.

When Should I Use an Expert?

While AI can help answer some questions about quality and regulatory requirements for medical devices, any answer generated by AI should be reviewed by an expert for accuracy. Bringing AI generated content to an expert for review can be a method to reduce the work required by an expert. But, be aware that if the AI generated content is poor quality, it can take longer to revise the AI generated content than it would have taken for the expert to create the content from scratch.

Do You Have Questions About When to Use AI?

Contact Cooperative Quality Solutions to discuss how to incorporate AI into your workflows while ensuring that any AI generated content is correct and compliant.