The health app I designed six years too early.
A university information-design brief: visualise health data. What I designed was an aggregator that pulls in your existing apps and devices, then displays everything on one canvas that changes mode rather than a stack of pages. Colour-coded domains, animated anatomical models, a rotatable muscle-recovery body, and a target version of yourself. In December 2019 that was speculative. Most of it now ships.
Visualise health data. That is the whole brief.
The module was information design, and the assignment was deliberately open: use colour, symbols, type and imagery to make deliberate aesthetic and cognitive choices that communicate effectively. No client, no platform, no constraints beyond the discipline itself.
A health app which integrates your existing app data in combination with your favourite smart devices to effectively visualise your progress and current status.
The framing in that sentence is the whole reason the project aged well. I did not set out to build another tracker. I set out to solve the fact that the data already existed and was scattered across half a dozen apps that refused to talk to each other, which is exactly the problem the category spent the following five years consolidating around.
Delete the navigation. Change the mode instead.
The first version was conventional: a navigation bar, buttons, and a separate page per visualisation. It was working, and it was wrong. The rewrite collapsed everything onto a single canvas that switches mode, so the interface never takes you anywhere, it just becomes about something else.
By putting my design on one page it allowed my design to flow, instead of a confusing journey.
That is a lazier and better answer, and it came from thinking about the user rather than the screen: someone mid-workout, phone in one hand, checking something in three seconds. A navigation hierarchy is a cost you pay for depth you do not need in that moment. Modes are what watches, rings and straps all eventually landed on.
Six domains. Six colours. One surface.
Each domain of health owns a colour, and selecting a mode floods the whole canvas with it. A saturated shade means active; a lighter shade means selected. You always know what you are looking at before you have read a single word.
- A 3D heart whose colour, size and animation encode the load being put on the organ
- A rotatable muscle body, each group shaded darker the more recently it was trained
- A food pyramid customisable per diet, where a good day resolves into a perfect shape
- A body you build from sliders, plus a target version of yourself as motivation
- An anatomical form is legible in a glance; a line graph needs reading
- The shape carries the meaning, so the numbers can stay secondary
- Graphs are still there underneath, one tap down, for anyone who wants them
- And it is honest about scale: heart data on minutes, workout data on weeks
What the category did next.
I am not claiming anyone copied a second-year student. The point is narrower and more useful: the reasoning was sound enough that the market independently arrived at most of the same answers, which is the strongest available evidence that the reasoning was right.
Pull from existing apps and devices rather than asking users to switch. Now the default assumption for every serious health platform.
A domain owns a colour, and the whole surface adopts it. Apple's activity rings and Whoop's strain and recovery language work the same way.
Modes rather than pages. Every watch face and ring app converged on a single canvas that changes what it is about.
Muscle-recovery models shaded by recency, and animated biometric forms, are now standard in training apps.
The one that is still genuinely unshipped is the target self: a personalised body built from your own measurements, held beside a version of the body you are working toward. It is the idea I rate most and the one I would be slowest to ship, and section 05 explains why.
Aggregating everything onto one surface is now the defining move of the category: Whoop Advanced Labs, Oura's lab uploads, Ultrahuman's Vision Cloud, Function Health, and Google's rebuild of Fitbit into the Health app. Colour-coded domain scoring is near-universal, Whoop's green, yellow and red recovery being the obvious case. Body composition arrived too, through Withings and Samsung's scales rather than the wearables themselves.
The target self is the exception. Amazon Halo shipped the closest thing in 2020, a 3D body model with a slider that morphed it across body-fat percentages, and nothing mainstream has shipped one since. The tell is Whoop Age: a target state expressed deliberately as a number and a colour rather than as a body.
What I would tear out.
- The aggregator thesis, and the discipline of not asking users to change behaviour
- Modes over pages, argued from the user's actual moment rather than from taste
- The colour system, which is doing real navigational work rather than decoration
- Interrogating the brief in writing before designing anything
- The BMI dial. BMI is a poor individual measure and I used it because it was familiar
- The target body shipped with no screening, no framing and no way to turn it off. Amazon's version drew sustained criticism on body-dysmorphia grounds, and at least floored its slider at a healthy body-fat percentage; mine had no such guardrail. The research on tracking and disordered eating is consistent but correlational, and there is none at all on 3D body visualisation specifically, so the honest objection is not "this is harmful", it is that I shipped a high-variance feature unguarded at 20
- Six modes is at least one too many; sleep and activity want merging
- No accessibility work at all, a colour-coded system with no secondary cue fails colour-blind users outright
This was graded on information design, not on product ethics or accessibility, and it shows. The visual reasoning was well ahead of where I was on responsibility, which is roughly the opposite balance to how I would weight the same brief now.