What does FDA · GMLP cover?

Good Machine Learning Practice. It applies to: Medical-device AI.

What does the Regulayer pack for FDA · GMLP prove?

Human-AI team performance and deployed-model monitoring for medical-device AI. The record is signed and tamper-evident, and anyone can verify it free, without an account.

Regulayer / The 146 laws / Pharma & life sciences

FDA · GMLP

Good Machine Learning Practice.

Instrument: Guiding principles, nonbindingApplies to: Medical-device AI.

What the signed record shows

Human-AI team performance and deployed-model monitoring for medical-device AI.

The proof is a signed, tamper-evident record. Anyone can check it, free, without an account, and nothing has to leave the building to make it. Evidence, not a promise.

Citation: GMLP Principles 1/2/7/9/10

In the same family

FDA · 21 CFR Part 11Electronic records & signatures EU GMP · Annex 11Computerised-systems integrity FDA-EMA · Good AI Practice (Jan 2026)Data governance, performance, life-cycle management

What to do about it

Keep the evidence this asks for, as the work happens.

A consequential AI-proposed action is checked against the human authority in force at that moment, and the decision leaves a signed record that a third party can verify independently, offline, without Regulayer. It runs inside your own environment and nothing has to leave it.

Life sciences: how the control works Or verify a real record, free →

Part of the Regulayer proof catalogue: 146 laws and standards, one sealed engine. This page is a product description, not legal advice.

What the statute requires, section by section

U.S. FDA, Health Canada, and U.K. MHRA, "Good Machine Learning Practice for Medical Device Development: Guiding Principles" (October 2021), ten guiding principles for the development of safe, effective, and high-quality AI/ML-enabled medical devices. This mapper addresses the principles that the deployment-time signed record can evidence:

Principle 1: Multi-disciplinary expertise leveraged throughout the total product life cycle (TPLC).
Principle 2: Good software engineering and security practices.
Principle 7: Focus on the performance of the Human-AI team.
Principle 9: Users are provided clear, essential information.
Principle 10: Deployed models are monitored for performance and re-training risks are managed. The development/validation-centric principles (3 representative data, 4 train/ test independence, 5 reference datasets, 6 model design, 8 clinical-condition testing) are pre-deployment artefacts outside the per-decision signed record.

Taken from the Regulayer entry for this instrument, which is built against the primary text.