What I did. As the designer, I tried to design a distillation: something that communicated severity, not just "normal vs. abnormal," across various kinds of lab data (numbers, yes/no results, ranges, related sub-labs). I proposed a "LabOrderBundle" concept and worked through the data model with engineering, then designed progressive disclosure so someone could get a quick "level with me, doc" read as well as dig deeper if they wanted.
How I know it worked. Going in, I had two goals:
- make the results readable at a glance
- actually explain what they mean instead of handing someone just the raw lab data
In unmoderated testing, people reflected those goals right back to me.
"I love the layout, it tells you straightforwardly what's abnormal and what's normal. Usually I just see the results and that's it."
"It tells me what it is, what the result means. My current program will tell me if something's normal or abnormal, but it won't educate me on what that lab is or what it's responsible for."
That was the signal it worked: testers describing the design intent back to me in the same terms I'd set out to hit.