Global Regulators publish Good Machine Learning Practices (GMLP)

by | Nov 1, 2021 | AI, Analytics, Compliance, Consulting Group, Data, FDA, Guidance, ISO, Medical Devices, MedTech, QMS, Quality, Quality Systems, Regulatory, Requirements, Robotics, SaMD, Standardization, Standards


It is no secret that AI/ML medical devices are under a special spotlight by international regulators this year. With more and more AI/ML SaMD’s being engineered, regulatory authorities like the FDA have been pushing for a more robust regulatory strategy. Last week, the FDA, Health Canada and the UK’s Medicines and Healthcare Products Regulatory Agency (MHRA) jointly published “10 Guiding Principles” for Good Machine Learning Practices (GMLP)[1]. These principles will lay the foundation for the newest branch of GxP’s aimed at developing safe and effective AI/ML medical devices.

Many of the foundational principles rely on good software engineering, such as data quality assurance, cybersecurity practices, data authenticity/integrity, and risk management. Additionally, the principles outline that the model’s intended integration into clinical workflow, and the benefits/risks should be leveraged throughout the total product life cycle. The following are additional principles outlined in the informal guidance:

  • Clinical study participants and data sets are representative of the intended patient population
  • Training data sets are independent of test sets
  • Selected reference datasets are based upon best available methods
  • Model design is tailored to the available data and reflects the intended use of the device
  • Focus is placed on the performance of the Human-AI team
  • Testing demonstrates device performance during clinically relevant conditions
  • Users are provided clear and essential information
  • Deployed models are monitored for performance, similar to post-market surveillance1

If you need assistance developing a medical device integrated with AI/ML, EMMA International’s team of experts can help with everything from a regulatory strategy, to developing software specification documentation, post-market vigilance, and beyond. Give us a call at 248-987-4497 or email to find out more!

[1] FDA (2021) Good Machine Learning Practice for Medical Device Development: Guiding Principles retrieved on 10/31/2021 from:

Madison Wheeler

Madison Wheeler

Director of Technical Operations - Ms. Wheeler serves as EMMA International’s Director of Technical Operations. She has experience in technical writing, nonconforming product management, issue evaluations, and implementing corrective and preventative actions in the pharmaceuticals and medical device industries. She has experience cross-functionally between R&D, lean manufacturing operations, and RA compliance. Ms. Wheeler also has academic and work experience with human health-risk engineering controls, physiological biophysics, and clinical research. Ms. Wheeler holds a Bachelor of Science in Biosystems Engineering with a concentration in Biomedical Engineering from Michigan State University. She is also a Certified Quality Auditor (CQA), and is currently pursuing her M.S. in Quality Management.

More Resources

Abbreviated 510k submission

Abbreviated 510k submission

There are three types of 510K, Premarket Notifications, which can be submitted to the Food and Drug Administration (FDA) traditional, abbreviated, and special. Abbreviated and Special 510K submissions can be utilized when the submissions meet the certain factors presented by the FDA. When submitting an abbreviated 510K the submission must include the elements that are identified in 21CFR 807.87 for the information required in a premarket notification submission.
Is your product a medical device?

Is your product a medical device?

Many marketed products are classified as medical devices and you would not even know it. Medical devices range from latex gloves and tongue depressors to respirators and heart valves. To determine if the product is considered a medical device by the Food and Drug Administration (FDA) you will need to analyze if your product meets the definition of a medical device per the Food, Drug, and Cosmetic Act1.
Sterilization of Medical Devices

Sterilization of Medical Devices

There are many ways to sterilize medical devices before use in the industry. The common methods utilized are steam, radiation, dry heat, ethylene oxide, and vaporized hydrogen peroxide. There are two methods of sterilization for medical devices recognized by the Food and Drug Administration (FDA), established and novel.

Ready to learn more about working with us?

Pin It on Pinterest

Share This