Life Sciences / Regulatory Brief 🧬

📌 Navigate

📊 Exec Summary

The February 9-15 window arrived one week after the most consequential device quality regulatory shift in three decades — QMSR's February 2 effective date — while the EMA-FDA joint AI principles published in January moved from announcement to active compliance planning context; together they define the infrastructure every MedTech and SaMD builder now operates inside.

Five things moved in life sciences / regulatory in the February 9-15 window:

The pattern: Regulatory infrastructure is not catching up to AI — it is getting ahead of it, and builders who map the new inspection, submission, and evidence frameworks now will have structural advantage over those who don't.

1️⃣ QMSR Live: FDA Retires QSIT, New Inspection Protocol

TL;DR: The Quality Management System Regulation took effect February 2, 2026, replacing 21 CFR Part 820 (QSR) by incorporating ISO 13485:2016 by reference — and FDA simultaneously retired the QSIT inspection technique and adopted a new protocol, resetting the inspection surface for every finished device manufacturer.

What happened

• QMSR effective date: February 2, 2026
• Replaces: 21 CFR Part 820 (the QSR, unchanged since 1996)
• Incorporates by reference: ISO 13485:2016 and ISO 9000:2015 Clause 3
• Inspection change: FDA retired QSIT (Quality System Inspection Technique) and adopted Compliance Program 7382.850
• Risk integration: risk-based thinking now required across operations, supplier management, software validation, production controls, change management, CAPA, and management review — previously confined to design controls only
• Enforcement scope: FDA investigators may review QMS records created before February 2, 2026
• Applicability: all finished device manufacturers commercially distributing in the US

📊 Key facts (from FDA / Ropes & Gray)

🔗 Primary source → QMSR State of Mind: FDA Adopts New Inspection Approach — Ropes & Gray

🔍 The non-obvious point

The new inspection protocol (CP 7382.850) is the operational unknown — FDA investigators are adapting in real time alongside manufacturers, which creates short-term inspection variability and long-term precedent-setting.

• Manufacturers already certified to ISO 13485 have the smallest compliance gap — QMSR is effectively the US finally catching up to the global standard
• The explicit extension of risk-based thinking to software validation and CAPA has direct implications for AI-enabled device software — these functions can no longer be treated separately from the QMS risk framework
• Records created before Feb 2 are in scope — retrospective audit exposure is real for teams that treated the old QSR as the ceiling

👀 What to watch

• First FDA 483 observations and Warning Letters issued under CP 7382.850 — will establish what the new inspection protocol prioritizes in practice

2️⃣ EMA-FDA Joint AI Principles Enter Active Compliance Context

TL;DR: The 10 joint EMA-FDA principles for good AI practice in drug development, published January 14, are now the operative global baseline for AI submissions — not prescriptive requirements yet, but the advance checklist every developer should map against.

What happened

• Published: January 14, 2026 (joint EMA-FDA)
• Scope: AI use across the full medicine lifecycle — early research, clinical trials, manufacturing, and pharmacovigilance
• 10 principles include: human-centric design, risk-based approach aligned to context of use, data governance, multidisciplinary expertise, lifecycle management, clear communication about AI systems
• Stakeholders: drug developers, marketing authorisation applicants and holders
• Status: not prescriptive requirements — intended to underpin future guidance documents in both FDA and EMA jurisdictions
• Process signal: EMA additional guidance expected Q2 2026

📊 Key facts (from EMA / FDA)

🔗 Primary source → EMA and FDA set common principles for AI in medicine development — EMA

🔍 The non-obvious point

The human-centric and lifecycle management principles are the ones most likely to show up as specific submission requirements in future FDA and EMA guidance documents.

• "Human-centric by design" is not a soft aspiration — it will translate into specific requirements around human oversight documentation, AI accountability frameworks, and fallback mechanism specs
• The lifecycle management principle signals that FDA and EMA expect ongoing evidence generation post-approval — not a one-time clearance event for AI-enabled drug products
• The EU AI Act's August 2, 2026 high-risk provisions create a hard compliance clock that intersects with these principles for any company with regulatory submissions touching EU markets

👀 What to watch

• EMA Q2 2026 guidance document — will be the first jurisdiction to convert these principles into specific submission expectations; FDA will likely follow within 6-12 months

3️⃣ FDA TEMPO Pilot: Reimbursement-Review Coupling for Digital Health

TL;DR: FDA launched the TEMPO pilot in cooperation with CMS, creating a limited pathway for digital health devices — including SaMD — to reach Medicare and Medicaid patients while FDA reviews safety and effectiveness; this is the most structurally significant SaMD go-to-market development in the Jan 2026 guidance package.

What happened

• TEMPO pilot launched as part of FDA's January 6, 2026 digital health guidance update
• Joint FDA-CMS mechanism: coordinates regulatory review with Medicare/Medicaid coverage determination
• Target: digital health devices seeking both clearance and reimbursement for government payer populations
• Context: FDA simultaneously updated general wellness and CDS software guidance (January 6), relaxing oversight on low-risk devices while creating structured pathways for higher-complexity SaMD

📊 Key facts (from FDA / Faegre Drinker / RAPS)

🔗 Primary source → Key Updates in FDA's 2026 General Wellness and CDS Guidance — Faegre Drinker

🔍 The non-obvious point

The TEMPO pilot is a reimbursement-review coupling mechanism — a limited structural change that reshapes go-to-market for SaMD builders, not just their regulatory pathway.

• For SaMD companies: TEMPO creates a limited regulatory-payer engagement rather than a universal sequential process (clear first, then fight for reimbursement separately)
• The Medicare/Medicaid population is the highest-complexity, highest-need target for many digital health devices — TEMPO narrows the reimbursement barrier that has historically caused cleared devices to stall commercially
• This is analogous to what the Breakthrough Device pathway did for faster review — TEMPO may do for faster payment coverage

👀 What to watch

• First TEMPO pilot participants announced — will reveal which device categories FDA and CMS prioritize for the coupled review process

4️⃣ PCCP: AI Device Updates Without New 510(k) Submissions

TL;DR: The FDA's final Predetermined Change Control Plan (PCCP) guidance is operative as of W07 — the mechanism exists and the framework is established, but industry practice around filing and approving PCCPs is still being built. Public precedent remains limited, and the practical iteration benefit depends on what FDA accepts in submitted change envelopes.

What happened

• PCCP guidance: final, issued August 2025, operative entering 2026
• Mechanism: manufacturers define covered changes upfront in the PCCP; FDA approves the plan; covered updates can then be deployed without a new 510(k), De Novo, or PMA submission
• Scope: AI-enabled device software functions
• Applicable pathways: 510(k), De Novo, PMA

📊 Key facts (from FDA / Ballard Spahr)

🔗 Primary source → FDA Issues Guidance on AI for Medical Devices — Ballard Spahr

🔍 The non-obvious point

PCCP is the most builder-relevant FDA AI mechanism right now because it converts a sequential gate (clear, then resubmit for every update) into a parallel architecture (define your change envelope, then iterate within it). That conversion is established regulatory policy — but the operational question of how broadly FDA will accept change envelopes is still being determined through submitted applications.

• Current expectation: PCCP guidance is final and operative; manufacturers can include a PCCP in their 510(k)/De Novo/PMA submission; FDA will review and approve the plan as part of clearance.
• Still-building practice: Public precedent remains limited on how broadly FDA will accept scoped change envelopes in practice; the first round of FDA feedback on submitted PCCPs will help establish the operational ceiling.
• For SaMD builders running continuous learning models: this means regulatory compliance can be designed into the MLOps pipeline, not bolted on as a stop-the-line event — but the scope of that design space is not yet settled.
• QMSR's new risk-based software validation requirements (effective Feb 2) interact directly with PCCP: the evidence standards for demonstrating that a covered change is within the approved envelope are now part of the QMS risk framework.

👀 What to watch

• First FDA feedback on submitted PCCPs — will set precedent on how broadly change envelopes can be scoped and what evidence FDA expects for covered changes; this feedback is the missing piece that converts the guidance framework into operational practice

5️⃣ AI Drug Pipeline: 173 Programs, Phase III Readouts in 2026

TL;DR: With 173 active AI-designed drug programs and the most advanced entering pivotal Phase III trials in 2026, this year will deliver the first at-scale validation data on whether AI-designed therapeutics perform at the level required to trigger regulatory submissions.

What happened

• 173 active AI-designed drug programs as of early 2026
• Most advanced programs entering pivotal Phase III trials in 2026
• FDA 7-step credibility framework operative for AI submissions — governs how AI-generated data is evaluated in regulatory filings
• EU AI Act high-risk provisions effective August 2, 2026 — may classify certain drug development AI as high-risk, adding compliance requirements for EU-market submissions
• Positive Phase III data triggers regulatory submissions; approval timelines extend to 2027-2028
• EMA guidance expected Q2 2026 — will align with FDA 7-step framework or create divergence

📊 Key facts (from Axis Intelligence / Drug Target Review)

🔗 Primary source → AI Drug Discovery 2026: 173 Programs — Axis Intelligence

🔍 The non-obvious point

The 2026 Phase III readouts are the first high-stakes test of whether AI-designed drugs perform at pivotal-trial scale — not just preclinically or in Phase I/II. The regulatory implications run both directions.

• Positive data: validates AI-designed drug as a regulatory submission pathway and sets precedent for how FDA and EMA evaluate AI provenance in submissions
• Negative data: does not invalidate AI drug discovery broadly, but will sharpen regulatory scrutiny on the credibility framework evidence standards required for future submissions
• The EU AI Act August 2 deadline creates a parallel pressure: companies with AI-designed programs targeting EU markets must map their AI tools against the high-risk classification criteria before Phase III data arrives

👀 What to watch

• EMA Q2 2026 guidance on AI in drug development — will determine whether EU submitters face additional evidence requirements beyond FDA's 7-step framework

📊 The pattern

Regulatory infrastructure moved faster than product in the February 9-15 window. QMSR replaced a 30-year quality regime on February 2. EMA and FDA established the global AI principles baseline in January. The TEMPO pilot introduced a limited FDA/CMS-linked pilot structure for SaMD. PCCP unlocked continuous AI device iteration. And 173 AI drug programs are closing on the first pivotal validation moment. The pattern is convergent: every major regulatory shift this window reduces friction for AI-enabled health products that plan submissions correctly and increases friction for those that don't. The regulatory stack is not a gate — it's becoming a runway.

👀 Watchlist

📎 Sources

Sources of truth

Also consider reading