Case Study · IMeasureU / Vicon · 2022–2024
A post-surgery rehab wearable started getting flagged for unreliable data. The hardware was fine. The UI was burying the one action that kept the data trustworthy — so users skipped it. This is how that got fixed.
In a nutshell
Knee Monitor is a system of wearable sensors and a companion app that helps physiotherapists accurately assess, monitor, and improve knee rehabilitation for patients recovering from surgery.
The accuracy of every measurement depends on one thing: the sensors staying in the right position on the patient's leg.
The situation
of users were complaining that the product wasn't telling the truth.
Our lab tests demonstrated very stable, reliable results. Out in the world, customers were saying something different.
Being an advocate for the best user experience, I stepped up to investigate the gap between what the device measured in a controlled environment and what physiotherapists were seeing in clinic.
Challenge overview
Sensor accuracy depends on proper alignment, and the sensors are prone to slipping or shifting during patient movement. Hardware improvements to the attachment method were slow. Algorithmic improvements were even slower. So the fastest path to better data ran through the interface.
The job was to make the need for re-alignment unmissable, and the re-alignment itself smooth enough that physiotherapists would actually do it mid-assessment.
Zero. Not "low adoption" — zero. The feature existed in the app and no one used it. That's the number that defined the brief.
Diagnosis
The re-alignment control sat at the top of the screen. Once the physiotherapist scrolled down to actually run an assessment, the button was out of view. The warning above it was easy to miss — and even when it wasn't, the path to act on it was painful:
Stack those three frictions on top of a clinician already managing a patient and a stopwatch, and the rational move is: skip it.
The team
A small cross-functional group with the Product Specialist embedded as a live bridge to the user community — he's a practising physiotherapist himself.
Approach
My priority as design lead was to make sure the team was building toward an outcome — better data — rather than shipping a feature. That meant getting close to the users, keeping engineering in the conversation early, and iterating in the open.
The PM and I embedded with physiotherapists — in-person interviews and observational sessions in real clinic settings. Workflows, frustrations, aspirations. Every design decision had a thread back to one of those rooms.
Our Product Specialist — a practising physiotherapist — was the live bridge to the user community. The Software Developer and Lead Algorithm Engineer were in early and often, so the constraint-feasibility conversation happened alongside the design exploration, not after it.
Figma for rapid prototyping. Usability sessions with physiotherapists for each pass. The feedback loop was tight enough that decisions had evidence behind them by the time they reached engineering.
Slack, Google Meet, and Zoom for the day-to-day. Miro and FigJam as the shared canvases for synthesis. Google Forms for short user pulses when we needed signal fast.
Process at a high level
Frame the re-alignment gap and its impact on data accuracy and trust.
Observe physiotherapists in real sessions. Interviews and surveys to surface pain points around sensor alignment.
Brainstorm approaches against research findings. Weigh feasibility and UX impact.
Translate the strongest options into interactive prototypes. Validate with physiotherapists.
Detail the final design, hand off to engineering, document outcomes and lessons.
The solution
A persistent tab attached to every test tile that requires sensors. It travels with the assessment, asks a single clear question, and demands no action unless the sensors actually moved.
Tab over button. A/B testing showed clinicians preferred the tab pattern over a button. Tabs suggest persistence and exploration — they let users investigate information without losing the context of the assessment they're running. The decision came from users, not from us.
User flow
During an assessment, the physiotherapist notices a sensor has shifted from its place.
"Has the sensor moved 1 inch or more?" — short, specific, impossible to misread. The physiotherapist taps the tab.
The re-alignment journey is incorporated into the test tile itself — no jumping to a separate screen and no lost context.
Once the re-alignment is done, the app auto-returns to the test tab the clinician was working in.
Supporting detail
Information about why alignment matters and when it's required is parked on an additional info tab — available when the user wants it, never blocking the assessment.
The results
The persistent, integrated re-alignment reminder addressed the issue on three fronts: it prompted users clearly without interrupting them, it dropped the cost of acting on the prompt to almost nothing, and it made data accuracy a natural side-effect of the workflow rather than a discipline the user had to enforce.
From 0% running the flow to 100% running it — and the trust signal in customer feedback moved with it. The hardware hadn't changed; the workflow had.
Lessons
In any system that relies on data for decisions, the interface has to actively prevent or mitigate sources of error — not just expose them and hope.
New features should feel like a natural extension of the user's normal flow. If your fix demands a context switch, it won't survive contact with a busy clinic.
Don't ask the user to spot problems. The system should surface them — and route the user toward the fix in the same gesture.
The tab-over-button decision came from users. The best calls in this project all came from listening to what the people in clinic were already telling us.