How Hackensack Meridian Health is building chronic care that reaches patients between visits. What health systems can learn from Hackensack about extending capacity and earning physician buy-in to expand access to care patients can count on. Hackensack Meridian Health's employed medical group delivers roughly 2.5 million visits a year across 450 care sites. Now the New Jersey-based system is setting a bold new standard for patient access and redesigning how it delivers care to meet it. At the Health Management Academy Leadership Summit earlier this October, Dr. Daniel Varga, President of the Physician Services Division and Chief Physician Executive at Hackensack Meridian Health, joined Cadence Founder and CEO Chris Altchek to discuss how the system innovated its chronic care model to meet the access standards the system has set. Through its partnership with Cadence, Hackensack is extending care into the homes of more than 6,000 enrolled patients, where connected devices and a clinical team help manage chronic conditions under the system's own protocols. Read on to see how Hackensack partnered with Cadence to reach its access goals, and what other health systems can learn from its approach to caring for patients between visits. Extending their statewide platform. Hackensack is a rapidly growing health system, expanding from roughly 1,000 physicians in 2018 to nearly 3,000 clinicians today, with more than 450 sites of care across New Jersey. With over half of Hackensack's 450 sites having five or fewer exam rooms, their existing demand kept their rooms full. To figure out how to increase capacity to better meet patients, Hackensack chose Cadence to extend the capacity of the system to deliver chronic care between visits. "Logistically, we just can't do this on our own," Dr. Varga said. Remote patient monitoring, chronic care management, and tech-enabled CMS models were new territory for the system, and Cadence offered a way to build them. Dr. Varga described the Cadence program as "a new way to extend care," and one that brought new funds rather than new cost. Remote monitoring can handle much of the routine management of diabetes, heart failure, and hypertension; freeing up space and time for Hackensack's clinicians to focus on seeing new patients and complex cases, while ensuring the full patient population is still receiving the highest quality care. What builds adoption. Offering a new tool did not ensure that Hackensack's clinicians utilized it. For Dr. Varga, adoption comes down to whether a program solves a problem for physicians - and what won physicians over was relief from the inbox. A patient message about swinging blood sugars or rising weight often ended with "why don't you come in?" Now a Cadence clinician adjusts insulin or a diuretic under protocol, and the physician hears about the result. As Dr. Varga put it, "Not having somebody come in or call or fill their inbox solved a big problem." And he's seen that smaller practices benefit too. "We actually have one-and-two-doctor practices who are big fans," Dr. Varga said. For a practice with two providers and four exam rooms, cutting a few unnecessary walk-ins and moving a patient from four follow-ups a year to three can open room for new patients. It's also more convenient for patients who can receive more care from home. Results from the program in practice. At 6,000 patients, initial clinical data is promising with a 76% relative increase in patients at <130/80 goal blood pressure after 6 months, and anecdotal stories about patients staying out of the ED. Altchek shared peer-reviewed data from Mayo Clinic Proceedings: Innovations, Quality & Outcomes, showing a 27% reduction in hospital admissions, concentrated in stroke, sepsis, cardiac dysrhythmia, and heart failure, conditions with long stays. He also highlighted that patients who are admitted have shorter stays on average. Altchek illustrated the results of the program with a recent example from a Texas health system. On a Saturday night, a patient entered a blood pressure reading of 190. Within three minutes, an AI voice agent and then a Cadence clinician called, coached the patient through taking additional medication, and asked for a recheck an hour later. When the reading came down, the patient was texted that no ED visit was needed, and a heart failure cardiology appointment was booked for Tuesday. He noted that sometimes the same monitoring might send a patient to the ED, but the real value is having a clinical team ready to assess and coordinate the right next step. The signal from physicians. The strongest endorsement, Altchek shared, has come from physicians themselves. Three health systems signed on after their physicians joined a different system and asked how they'd been practicing without it. While anecdotal, it points to where he views the industry heading: white-label partners delivering meaningful care under shared clinical protocols and payment arrangements with payers. The goal is "hopefully not a 20% increase in capacity, but a 100% increase in capacity." What would greater chronic care capacity make possible for your health system? Connect with Cadence to explore how its clinical team can help expand access under your protocols, support your physicians, and deliver care between visits.

FDA opens GenAI consultation while ISO 13485 timelines tighten. Sherif Elkhadem 19 August 2026 FDA just opened public consultation on generative AI in medical devices, the same week Greenlight Guru published updated ISO 13485 certification timelines showing the process now routinely exceeds 18 months. These are not unrelated events. As regulatory authorities grapple with rapid AI innovation, they are simultaneously tightening the baseline quality infrastructure manufacturers must demonstrate before any device reaches market. For regulatory affairs and quality teams, this dual signal reveals where the regulatory centre of gravity is moving. The FDA consultation, announced through a Request for Information, acknowledges that generative AI-enabled devices introduce unique risks compared to traditional software or even deterministic AI. The agency is seeking feedback on risk assessment, validation, transparency and post-market monitoring for devices that can generate novel outputs not explicitly programmed. Meanwhile, ISO 13485:2016 certification, the recognised foundation for medical device quality systems globally, now requires a timeline most startups dramatically underestimate. The intersection of these two developments is not coincidental. It is structural. Why FDA is asking now, and what the questions reveal. FDA's Request for Information on generative AI devices is notable less for what it asks than for when it asks. The agency already has a framework for AI and machine learning devices, updated in 2021 with the concept of predetermined change control plans. It already cleared multiple AI diagnostic tools under existing pathways. But generative AI presents a different challenge. These devices do not simply classify or detect. They produce content, recommendations or outputs that may not exist in training data. That fundamentally changes the risk profile. The consultation signals FDA is building a regulatory approach before widespread submissions arrive, not after. It follows a similar pattern to the MHRA's ambient voice technology guidance, where the regulator published classification criteria for a technology category at the point of clinical adoption, not market saturation. For manufacturers developing generative AI devices, this is a narrow window to shape regulatory expectations before they solidify. The questions FDA is asking, particularly around validation and transparency, suggest the agency is looking for evidence frameworks that extend well beyond traditional V&V protocols. Critically, the consultation arrives as FDA names participants in its TEMPO digital health pilot, a Medicare-linked pathway that exempts certain devices from some premarket requirements in exchange for real-world evidence collection. Cadence, the second participant announced this week, develops AI-enabled software for hypertension management. The juxtaposition is deliberate. FDA is signalling parallel tracks: streamlined pathways for devices with established risk profiles, and heightened scrutiny for generative AI where risk models are still forming. ISO 13485 timelines stretch as certification bodies tighten scrutiny. Greenlight Guru's updated ISO 13485 certification timeline reveals a process that now routinely takes 18 months or more from gap analysis to certificate issuance. This is not a bureaucratic delay. It reflects certification bodies conducting deeper audits, particularly around risk management, design control traceability and post-market surveillance integration. For startups approaching their first certification, the timeline represents a funding and planning challenge that compounds if not anticipated early. The timeline is driven by several structural factors. Notified bodies and certification bodies are under heightened regulatory scrutiny following MDR implementation in Europe and ongoing FDA collaboration through mutual recognition agreements. Auditors are spending more time on evidence architecture, particularly the linkage between risk assessments, design inputs, verification outputs and post-market data. They are also requiring more robust supplier management documentation, especially for SaMD teams relying on third-party components or cloud infrastructure. As LLM-enabled SaMD development becomes common, supplier qualification is a growing audit flashpoint. The extended timeline has knock-on effects for regulatory submissions. Many jurisdictions, including Canada under MDSAP and Australia under TGA, accept ISO 13485 certification as part of the conformity assessment route. Without the certificate in hand, market entry timelines stretch further. For manufacturers targeting multiple markets simultaneously, as RevealDx recently demonstrated with its triple-clearance AI lung nodule tool, ISO 13485 certification becomes the long pole in the schedule. Where generative AI and quality systems intersect. The FDA consultation and ISO 13485 timeline pressure converge on a single compliance question: how do you validate and maintain a quality system for a device that produces outputs you cannot fully enumerate at the point of design freeze? Traditional design control processes assume a defined specification. You design to it, verify against it, validate in intended use, then lock it. Generative AI devices challenge every step of that sequence. ISO 13485 requires manufacturers to define acceptance criteria for design outputs and maintain traceability between design inputs, risk controls and verification results. For a generative AI device, those acceptance criteria must account for stochastic behaviour, edge cases that emerge post-deployment and model drift over time. That means quality systems must integrate continuous monitoring, version control for model retraining and a change control process that can distinguish between acceptable variation and a change requiring regulatory notification. This is not a theoretical challenge. Aidoc's generative AI radiology tool, submitted to FDA earlier this year, required a validation strategy that extended beyond traditional sensitivity and specificity metrics. The company had to demonstrate how the system's generative outputs would be bounded, how clinical users would be alerted to uncertainty and how post-market data would feed back into model governance. That level of integration between algorithm development and QMS infrastructure is not standard in most ISO 13485 systems today. What this means for your team. If your device pipeline includes any form of generative AI, the FDA consultation window is a strategic opportunity. The Request for Information is open for comment, and early input has historically influenced how FDA structures subsequent guidance. This is the moment to surface your validation challenges, evidence gaps and operational constraints before the agency formalises its expectations. Once guidance is published, the compliance baseline is set. For teams approaching ISO 13485 certification, the 18-month timeline is no longer an outlier. It is the median. That means certification planning must begin well before first submission, ideally during design and development phases when QMS integration is least disruptive. Waiting until a device is feature-complete to build a quality system around it is a recipe for non-conformities, rework and timeline slippage. Certification bodies are increasingly focused on whether the QMS was used to develop the device, not simply documented after the fact. The intersection of these timelines matters most for startups targeting multiple markets. If your strategy includes FDA clearance, CE Mark under MDR and markets requiring ISO 13485 certification, the certification timeline becomes your critical path. You cannot compress it meaningfully once you start. That makes early investment in QMS infrastructure a funding and operational priority, not a compliance overhead to defer. Key takeaways. * FDA's generative AI consultation signals the agency is building regulatory expectations before widespread submissions arrive, creating a narrow window for manufacturers to shape guidance through public comment. * ISO 13485 certification now routinely exceeds 18 months, driven by deeper audits on risk management, design control traceability and post-market surveillance integration. * Generative AI devices challenge traditional design control assumptions, requiring quality systems that integrate continuous monitoring, version control and change management from the outset. * For multi-market strategies, ISO 13485 certification is often the long pole in the timeline, making early QMS investment critical for funding and submission planning. * Certification bodies are scrutinising whether the QMS was used to develop the device or simply documented retrospectively, shifting audit focus from paperwork to evidence architecture. The regulatory environment for AI-enabled devices is moving from reactive to anticipatory. FDA is consulting on generative AI before the submission wave peaks. Certification bodies are auditing quality systems with an eye to post-market resilience, not just premarket compliance. For manufacturers, this shift demands earlier, deeper integration between product development and regulatory infrastructure. The teams that recognise this are building quality systems and evidence architectures in parallel with their algorithms, not sequentially. That is the competitive signal in this week's announcements.