"Human-centered" is one of the most-used phrases in medical technology — and one of the easiest to confuse with something much narrower. A device can pass a human-factors test and still have been centred on no one, because a test is not a centre.
To test a device against a human is to ask whether the human can survive the device. To centre a device on a human is to build the device around the human in the first place. One is a checkpoint. The other is a starting point. The two are easy to confuse, and the confusion is costly.
Article 001 named seven structural patterns by which the clinical environment rejects products the clinical problem said should succeed. This essay is about the discipline that is supposed to prevent all seven — and why, as most MedTech companies practise it, it does not.
The discipline has two names, and the distance between them is the whole point. Human factors is the study that gets run before launch to show a representative user can complete the task safely. Human-centered design is the practice of shaping the product around the actual human who will live with it. The first is often treated as if it satisfies the second. It does not, and the difference is not academic. It is the difference between a product that clears and a product that adopts.
01The claim, and what backs it
Open the marketing of a medical device and the language is familiar: patient-centred, clinician-centred, designed around the user. The intent behind it is usually real — the team means it. But the evidence that backs it is often a single artefact: a human-factors validation study, run once, before launch, to satisfy a regulatory requirement.
That study is real work and often rigorous. But notice what it is: a test of whether a representative user, under controlled conditions, can complete the critical tasks without dangerous error. It answers can the human use it safely. It says nothing about whether the product was built around the human, whether the median clinician will choose to use it on a busy day, or whether it earns its place in the work eighteen months in. It is possible to pass that study with a device that was never centred on anyone — and, in good faith, to call it human-centered.
This is not dishonesty. It is a category error, and it is expensive, because it lets a company believe the human question is answered when it has barely been asked.
02Tested is not centered
The gap between testing a human and centring on one comes down to two questions that sound similar and are not.
The test asks: can a representative user complete the task safely, under controlled conditions, before launch? Human-centered design asks a larger question that runs the whole life of the product: will the median clinician, at capacity, unaccompanied, keep choosing to use this the way it was intended — in the four-hundredth session as in the first — and does the product make their work better, not merely survivable?
The first is a gate. The second is a life. Measuring the gate and inferring the life is where products quietly die. Human-centered design is not a bigger validation study and it is not a compliance step done more thoroughly. It is a different act, performed at a different time: it happens before the device is finished, and it is drawn from the field, not the lab.
A human-factors test is done to the device. Human-centered design is done to the whole idea of the device — before there is a device to test.
The tell is timing. If the first serious contact with the real human happens in a validation room near launch, the company is human-testing. If it happens in the clinic before the concept is locked, the company is human-centring. Same word on the website. Opposite practice.
03Two humans, two rooms
The reason a passed test does not guarantee a centred product is that the human in the test is not the human in the field. They pass through two different rooms.
In the first room — validation, demonstration, the lead site — the device is the only thing happening. Attention is undivided. A trainer is present. The device is fresh, the user was selected, the clock is generous, and someone is watching to make sure it goes well. Under these conditions the device performs as designed, because the engineering is sound for this room.
In the second room — the median clinic on an ordinary day — the device is one demand among forty. Attention is fragmented. The user is at the end of a shift, no representative is present, the clock is real, and so is the patient. Nothing about the engineering has changed. Everything about the human has. A rested operator doing one thing and an overloaded operator doing forty are not the same user with a small correction between them; they are two different people, and the device meets both.
Human-centered design is the discipline of designing for the human in the second room. A human-factors study that observed only the first room studied the wrong human — carefully, expensively, and to the wrong end. This is why "we did our usability testing" and "our product isn't adopting" are not a contradiction. They are the ordinary result of centring on the first-room human and shipping to the second.
04Where the real human breaks
When a device meets the second-room human, it is not rejected dramatically. It is rejected through a series of small, rational decisions, each invisible on its own. Three of them account for most of the loss — and all three are decisions a human-centered process would have seen coming.
Break one · the loadThe capacity runs out
The device adds cognitive load — a screen to watch, an alert to read, a decision to make — at the exact phase when the human's load is already at its ceiling. The first-room human had spare capacity to absorb it. The second-room human does not, and so sheds load the only way available: using the device in the easy cases where there is room for it, and avoiding it in the hard ones it was built to justify. The value thesis does not fail loudly. It inverts, quietly, one rational choice at a time.
Break two · the bendThe work won't reorganise
To be used as designed, the device asks the human to reorganise the work around it. The first-room human accepted the ask; they were invested and had time. The second-room human runs the same trade in real time and declines — correctly, within their own frame. The device is entirely capable. The ask is simply larger than the benefit they can perceive from where they stand. Capable and declined are not a contradiction; they are what happens when the work was never centred in the design.
Break three · the strangerThe human who was never in the room
The human who was shown the device — the champion, trained and accompanied — is not the human who inherits it. The product was optimised, consciously or not, for the person the company kept meeting. The median human downstream was never in that room. Adoption reads as strong while the device is still meeting people like the champion, and plateaus the moment it reaches the humans it was never designed for — by which point the early enthusiasm has already been booked as proof.
05Design from the clinic
Human-centered design, taken at its word, is not a study you run and file. It is a map of the actual humans the device will meet — their load, their workflow, their incentives, their ordinary day — drawn from the clinic before the concept is locked. It does not replace the human-factors study; it precedes it and gives it something true to test against.
Done before the build, that map changes what gets built. The load that runs the user out of capacity is designed out. The ask the work will refuse is made smaller. The device is aimed at the median human, not the champion. Done before launch, it changes how the product is deployed and to whom. Done only after the plateau, it becomes triage — still worth doing, far more expensive, and working against a product that has already spent its early credibility.
The clinic is not where you test the design. The clinic is the design brief. A device centred anywhere else was centred on the wrong thing.
This is the reframe the sequence turns on. The seven patterns of Article 001 are not seven independent risks to manage after the fact. They are what a first-room design produces when it meets a second-room world. Centring the design on the real human, in the real clinic, before the concept is fixed, is the one intervention that acts on all seven at once — because it removes the gap they all grow in.
06Is your product human-centered?
The distinction is easy to claim and hard to pass — and the honest way to find out is to answer for your own product, specifically, not in the abstract. That is not a checklist you can skim. It is a short, structured self-test.
The self-testTen questions, sent on request
I've written ten questions that separate a human-centered product from a human-tested one — covering when the real human first shaped the design, whether the median user was ever watched at capacity, where the product runs the clinician out of capacity, and whether the human who was shown the device is the human who now lives with it. They work as a private self-assessment, or as the starting point for a conversation.
Contact me, and I'll send you the ten questions.
Whatever the answers turn out to be, the fix has the same shape: not more testing or a lower price, but going to the clinic and letting the real human shape the product before the next one is built.
That is what human-centered actually means. It begins in the clinic, not the report.