GMLP
Good Machine Learning Practice
What a definition is not
A definition is SPEQ’s plain-language decode of how a term is used in practice, cited to the documents that define it. It is a practitioner reference, not legal or regulatory advice, it does not replace the definition in the source, and where a regulator’s wording differs the regulator’s wording governs.
Good Machine Learning Practice is a set of guiding principles, issued jointly by FDA, Health Canada, and the UK MHRA, for the development of machine-learning-enabled medical devices — covering the full product lifecycle from data management and model training to evaluation, deployment, and monitoring. The principles emphasize representative and well-characterized data, robust independent evaluation, human factors, transparency to users, and monitoring of deployed performance. GMLP is high-level and principle-based, intended to guide practice rather than prescribe a single method.
The principles target the failure modes specific to ML: models that perform well on training data but degrade on real-world or shifted populations, opaque decision-making, and performance drift after deployment. They push developers toward representative datasets, clear separation of training and test data, and ongoing real-world performance monitoring.
GMLP sits alongside the emerging framework for AI in the wider GxP context — including the FDA/EMA guiding principles of good AI practice in drug development and predetermined change control plans for adaptive algorithms — as regulators converge on how to trust learning systems.
- —Joint FDA / Health Canada / MHRA guiding principles for ML-enabled devices
- —Covers the whole lifecycle: data, training, evaluation, deployment, monitoring
- —Emphasizes representative data, independent evaluation, transparency, drift monitoring
- —Principle-based guidance, not a prescriptive standard
FDA / Health Canada / MHRA "Good Machine Learning Practice for Medical Device Development: Guiding Principles".
Frequently asked questions
What does GMLP stand for?
GMLP stands for Good Machine Learning Practice.
What is GMLP?
Good Machine Learning Practice is a set of guiding principles, issued jointly by FDA, Health Canada, and the UK MHRA, for the development of machine-learning-enabled medical devices — covering the full product lifecycle from data management and model training to evaluation, deployment, and monitoring. The principles emphasize representative and well-characterized data, robust independent evaluation, human factors, transparency to users, and monitoring of deployed performance. GMLP is high-level and principle-based, intended to guide practice rather than prescribe a single method.
Which regulations cover GMLP?
FDA / Health Canada / MHRA "Good Machine Learning Practice for Medical Device Development: Guiding Principles".