Blog

HL7 International, the organization behind the FHIR standard, launched a new talent program this week aimed at a problem most MedTech companies feel directly but rarely see written down: there simply aren't enough people who know how to build FHIR integrations properly. The program, called Break the Code, is a free global initiative giving students hands-on experience building FHIR applications, starting this October in partnership with CAHIIM. Behind the announcement is a number worth sitting with: HL7's own workforce analysis estimates the global healthcare ecosystem could need more than 1 million additional skilled FHIR workers over the next five years.

If you've heard about the European Health Data Space, you probably filed it under "someone else's problem." Sounds like an EHR thing. A hospital IT thing. Not something that touches your medical device software. That's not quite right, and the gap between what people think this rule covers and what it actually covers is worth closing now, not in three years.

Medical device software should not only satisfy today's requirements. Its architecture should make future hardware, software, connectivity, and feature changes manageable. Early software architecture advisory can reduce technical risk, avoid expensive redesigns, and prepare a medical device for long-term development. The goal is to design for change, not only for the first release.

Since October 2023, FDA has had the authority to refuse a 510(k) outright if the cybersecurity documentation isn't there. Not delay it. Refuse to even review it. FDA tightened the requirements further with its June 2025 final guidance update, and heading into 2026, the same gaps keep showing up in submissions. We looked at what's actually driving those rejections, and the pattern is consistent across the industry.

The prototype worked. The clinical team was impressed. The investors nodded. And then, somewhere between "this works" and "this is live," everything stopped moving. Timelines slipped. Costs grew. Months passed. This is the most common story in healthtech. And it almost never happens because the idea was wrong.

On 14 September 2026 the FDA released a new educational course on patient-generated and real-world data across the medical device lifecycle. The course makes one point clear: a device’s usefulness doesn’t end once it reaches the market. How patients actually use it generates data that helps manufacturers see how the product performs and how it could evolve in the future.

Medtronic, one of the world's largest medical device companies, has announced a $700 million investment in Cornerstone Robotics. The agreement gives Medtronic access to Cornerstone's robotic surgery platform and distribution rights for its Sentire surgical system outside the United States. Rather than replacing Medtronic's own Hugo robotic-assisted surgery system, Sentire is expected to complement it. Together, the two platforms expand the options Medtronic can offer hospitals and surgeons within its connected surgical ecosystem.

MedTech innovations do not always require a new sensor or device. Hospitals and healthcare companies already generate large amounts of data, including ECG recordings, medical images, lab results, and patient monitoring data. However, this information is often located in different systems, stored in inconsistent formats, or used only for its original purpose. A recent example of how AI can uncover additional insights from data already collected is Tempus ECG-PH. It received FDA 510(k) clearance in August 2026. The software analyzes standard 12-lead ECG recordings for signs associated with pulmonary hypertension, instead of requiring a new type of device. This kind of approach is already emerging across the industry: AI models built on top of standard diagnostic recordings, such as a routine ECG, can be trained to flag signs of conditions the test was never originally designed to detect. No new sensor, no new device, just a new way of reading data that was already being collected.

On the 10th of August 2026, the FDA proposed reclassifying DBT (digital breast tomosynthesis) systems from Class III requiring PMA, to Class II with special controls and the 510(k) pathway. This article presents DBT as a case study of how an entire device category can evolve from a high-risk technology to a well-understood product class. It argues that a technology becomes more established from a regulatory perspective as evidence grows and its risks become better understood and easier to manage.

When an AI model drives a diagnosis or a treatment decision, it's not just a feature of the device. It's the core of what the device does. Regulators treat it that way, and that changes how you need to build it. You can't just train a model, get good accuracy, and then look for a regulatory pathway afterward. By the time you're writing the submission, it's too late to fix what should have been built in from the start.