FDA–MHRA Liaison and SaMD Evidence: What Just Changed

Two developments this week should reshape how regulatory teams approach software as a medical device (SaMD) submissions, particularly those targeting both US and UK markets. The MHRA announced a unique liaison programme with the FDA—the first of its kind—while a former FDA medical officer published guidance clarifying what continuous clinical evidence actually means for software devices. Taken together, they signal a fundamental recalibration: regulators are aligning procedurally while simultaneously raising evidentiary expectations post-market.
For manufacturers pursuing dual FDA–MHRA pathways, this isn't just diplomatic news. It's a strategic inflection point that will affect submission timing, clinical evaluation planning, and post-market surveillance architecture. Here's what changed, and what your team needs to do now.
The FDA–MHRA Liaison Programme: Beyond Symbolic Cooperation
The MHRA's announcement of a unique liaison programme with the FDA goes well beyond the usual bilateral pleasantries. This is a structured, formal exchange designed to embed regulatory personnel within each agency's review processes. The stated aim: reinforce collaboration, harmonise scientific assessment approaches, and accelerate mutual learning on emerging technologies—particularly AI-enabled devices and SaMD.
What makes this operationally significant is timing. It follows the MHRA's launch of an AI regulatory sandbox aimed at improving medicines safety and accelerating development timelines. The FDA, meanwhile, has been expanding internal AI capabilities and shifting toward continuous oversight models that favour real-world performance data over static pre-market submissions. The liaison programme effectively creates a regulatory back-channel for aligning these parallel modernisation efforts.
For regulatory teams, the implications are tactical. Expect greater consistency in how both agencies interpret software validation evidence, risk classification for adaptive algorithms, and post-market performance expectations. This doesn't mean identical requirements—the FDA's 510(k) and MHRA's conformity assessment routes remain structurally distinct—but it does mean fewer divergent interpretations on core evidentiary questions. If you're planning a SaMD submission to both markets, this liaison increases the value of early engagement with both agencies in parallel, rather than sequential filings.
It also raises the floor for what 'adequate' clinical evidence looks like. If both agencies are aligning on scientific assessment frameworks, manufacturers relying on minimal datasets or narrow validation cohorts will find themselves out of step with evolving expectations. This is where the second development becomes critical.
Continuous Clinical Evidence: What a Former FDA Medical Officer Says You're Getting Wrong
Dr. Bakul Patel, formerly a medical officer at FDA's Center for Devices and Radiological Health, published new guidance this week clarifying the single most misunderstood requirement for SaMD: continuous clinical evidence generation. His core observation—based on 12 years inside the FDA and 25 as a cardiothoracic surgeon—is blunt: most software device companies still treat clinical evidence as a pre-market hurdle, not an ongoing obligation.
The distinction matters enormously under both FDA and MHRA frameworks. For SaMD risk management under ISO 14971, clinical evidence isn't a static dossier compiled once and filed away. It's a living dataset that evolves as the software is deployed, updated, and used in real-world settings. This is especially true for adaptive algorithms, where performance can drift as input data distributions shift post-market.
Patel's guidance highlights three operational gaps that trip up manufacturers: **First**, many SaMD developers rely exclusively on retrospective validation datasets—often sourced from a single institution or narrow patient demographic—without planning for prospective, real-world performance monitoring. This is increasingly untenable as regulators demand evidence that software performs as intended across diverse clinical environments and patient populations. **Second**, post-market surveillance infrastructure is often bolted on as an afterthought, rather than designed into the product architecture from the start. Continuous evidence generation requires instrumentation: logging, performance metrics, and data pipelines capable of tracking device behaviour in production. If your SaMD can't systematically capture and report performance data post-deployment, you're not compliant with emerging expectations—you're running on borrowed time. **Third**, many teams underestimate the resourcing required for ongoing clinical evaluation. This isn't a one-time documentation exercise; it's a continuous function that requires clinical expertise, data infrastructure, and regulatory coordination. As clinical evaluation under EU MDR has demonstrated, regulators are scrutinising not just the volume of evidence, but its relevance, recency, and rigour.
For SaMD manufacturers, the takeaway is unambiguous: if your clinical evidence strategy stops at FDA clearance or MHRA conformity assessment, you're exposed. Post-market performance data is no longer a 'nice to have'—it's table stakes for regulatory sustainability and increasingly a differentiator in competitive landscapes where payers and clinicians demand real-world validation.
NuraLogix's 510(k) Clearance: A Case Study in Contactless Measurement
This week's FDA clearance of NuraLogix's Anura Mobile Core SDK for contactless measurement of heart rate and respiratory rate offers a concrete example of how these evidentiary standards are playing out in practice. The SDK uses smartphone cameras to derive physiological parameters—a novel measurement modality that required the company to demonstrate not just analytical validity, but clinical reliability across diverse skin tones, lighting conditions, and device hardware.
What's instructive here is less the technology itself than the regulatory pathway. NuraLogix pursued 510(k) clearance by establishing equivalence to predicate devices—a well-trodden route for Class II devices. But contactless measurement introduces variables (ambient lighting, camera quality, user motion) that don't exist with traditional contact-based sensors. The FDA's clearance implicitly validates the company's approach to controlling and documenting those variables, which will set a precedent for subsequent contactless diagnostic tools.
For SaMD teams working on novel measurement modalities—whether imaging analytics, wearable-derived biomarkers, or remote monitoring algorithms—NuraLogix's clearance underscores the importance of validation breadth. Demonstrating performance in controlled lab settings is necessary but insufficient. Regulators increasingly expect evidence that devices perform reliably under real-world variability: diverse patient populations, varied clinical workflows, and heterogeneous hardware ecosystems.
What This Means for Your Team
If you're developing SaMD for US and UK markets, these three developments should prompt immediate tactical adjustments. First, revisit your clinical evidence strategy. If it's anchored exclusively in pre-market validation, you're misaligned with where both FDA and MHRA oversight is heading. Plan for continuous evidence generation from day one: instrument your software to capture performance data, establish prospective monitoring protocols, and resource ongoing clinical evaluation as a permanent function, not a project milestone.
Second, treat the FDA–MHRA liaison programme as an opportunity to front-load regulatory alignment. If you're targeting both markets, early parallel engagement—via pre-submission meetings with FDA and MHRA consultations—can surface divergent expectations before they become submission deficiencies. The liaison programme increases the likelihood that scientific advice from one agency will inform the other's perspective, but only if you engage both early and transparently.
Third, for teams pursuing novel measurement modalities or adaptive algorithms, NuraLogix's clearance offers a template: validation breadth matters more than validation volume. Regulators want evidence across use conditions, not just within a single controlled cohort. This means designing clinical studies that deliberately stress-test your device against variability—diverse demographics, variable lighting, different hardware—and documenting how performance holds (or degrades) under those conditions.
Finally, recognise that continuous compliance is replacing pre-audit scrambles as the industry standard. Both FDA and MHRA are shifting toward continuous oversight models that favour manufacturers with robust, real-time quality and performance data. If your post-market surveillance infrastructure can't answer basic questions—'How is the device performing in production?' 'What adverse events have been reported?' 'How has performance changed since the last software update?'—you're not prepared for where regulatory scrutiny is going.
Key Takeaways
- The FDA–MHRA liaison programme signals procedural alignment on SaMD and AI devices, increasing the value of parallel early engagement with both agencies rather than sequential submissions.
- Continuous clinical evidence is now the regulatory expectation, not a post-market afterthought—instrument your SaMD to capture real-world performance data from deployment, not retrospectively.
- NuraLogix's 510(k) clearance for contactless measurement demonstrates that validation breadth across real-world variability (diverse demographics, lighting, hardware) is as critical as analytical accuracy.
- Post-market surveillance infrastructure must be designed into product architecture from the start—regulators increasingly expect real-time performance data, not annual summary reports compiled after the fact.
For regulatory teams navigating SaMD pathways, this week marks a quiet but consequential shift. The FDA and MHRA are converging on core scientific expectations while simultaneously raising the bar for post-market evidence. Manufacturers who treat clinical evaluation as a living discipline—rather than a compliance checkpoint—will find themselves better positioned not just for regulatory approval, but for sustained market access as oversight models continue to evolve. The question isn't whether your evidence is sufficient for clearance today, but whether your infrastructure can sustain the evidentiary expectations of tomorrow.