Key Takeaways
- User research for medical devices — not asking "what do you want," but watching "how do you work"
- One testing device from concept to T0 trial mold, through 5 major revisions and 3 complete restarts
- Three design pitfalls most overlooked in medical products: disinfection, cable routing, and center of gravity
- CMF's special rules in medical equipment — it's not about looks, it's about durability and cleanability
In the fall of 2024, we took on a medical testing device project. The client was a Shenzhen-based startup specializing in in-vitro diagnostics. Their first generation product had been on the market for two years, and the most common user feedback wasn't about accuracy — it was "hard to use." Honestly, that feedback got me excited — it meant the core product was solid, and the only thing missing was crossing the threshold from "functional" to "usable." I followed this project closely for over half a year, so let me walk through the whole process.
We Showed Up On-Site and Realized Everything We Thought Was Wrong
The client's initial brief was clear — "a desktop biochemical analyzer, with a strong tech look, not too big." Sounded pretty standard, right? But we insisted on doing user research first. To be honest, the client's expression at the time was basically "what's there to research, a hospital is a hospital."
I took the team to the clinical labs of two top-tier hospitals in Guangzhou, and we camped out for three full days. Would you believe it — on day one we found a critical flaw. The standard lab benchtop height is 75 cm, but the device's sample inlet was designed on top of the machine. A 165 cm tall nurse had to stand on her toes to read the screen and reach up to place samples. To be fair, that one's on the mechanical engineers — they designed the operating height from their own 180 cm perspective.
Even more absurd was the cable routing. The original device's power and data cables came out the back. On a cramped hospital benchtop, the cables either hung down to the floor where people stepped on them or looped around to the front blocking the display. The nurses told us they taped the cables under the table with duct tape — can you believe that's the solution being used in a hospital?
Five Versions, Three Restarts, Every One a Bloody Lesson
V1 was a "looks like a medical device" design — white enclosure, blue light strips, rounded corners. Honestly, the kind of design you could drop into any hospital brochure and it wouldn't look out of place. The client said "it's fine," but the more we looked at it, the more something felt off. Put this thing next to twenty-odd competitors on the market and patients couldn't tell them apart. Tore it up and started over.
V2 swung to the other extreme — a biomimetic style, trying to make the device look like an organic form "like a human cell." The client's boss loved it, said "it has identity." But the moment I talked to the mold maker, I knew we were in trouble. That double-curved surface required at least three-side sliders on the mold, ejection pin marks couldn't be hidden, and surface finishing alone would take two extra weeks. When the T0 trial mold quote came in, it was 40% higher than a conventional design. The proposal got killed — not by the client, but by me. Making something beautiful that can't be manufactured, or that doubles the cost to produce — that's an artist, not an industrial designer.
I won't go into detail on V3 through V5, but the trend was increasingly "restrained." The final version scrapped all curved forms, replacing them with planar segmentation and material contrast to create a sense of technology. The front used matte PC, the side panels used metallic grey sandblasting. The only curved surface was the angled sample inlet — and that angle was adjusted four times, from 15° down to 9°, to match the natural angle of a nurse's extended arm.
Three "Small Problems" Everyone Overlooked
Disinfection. Medical equipment gets wiped down with alcohol at least three times a day. You think any random surface finish can handle that? Not even close. At one point we chose a UV matte coating. The sample came out gorgeous — the feel was as smooth as baby skin. But after 200 wipes with an alcohol pad, the surface was already turning white and peeling. We ended up switching to textured grain + full-color injection molding — no matter how many times you wipe it, the plastic's natural color stays right there.
Cable routing. An electronic engineer who's never been on-site will never understand how cables actually need to run. I once saw the internal cable routing of an imported device — honestly, I was blown away. Every single cable was secured in cable management channels, with rounded corners at every turn, as well-maintained as a luxury interior. But with many of our domestic devices, pop open the bottom cover and you find a plate of spaghetti. This time we built dedicated cable channels and clip structures into the bottom housing — the routing path was designed right into the molded part. This item alone went through five revisions before we locked it down.
Center of gravity. The heaviest component inside this machine was the temperature control module, roughly 3 kg. In V3 we placed the module right in the center of the chassis — theoretically a perfect weight distribution. But when the T0 trial mold came back, we discovered that if a user pressed on the upper-right corner of the machine while operating the touchscreen, the whole unit would wobble slightly on the benchtop. We spent the next two months moving the temperature control module to a lower rear position, and simultaneously adjusting the height and durometer of the four rubber feet. Now you can stand next to the machine and push down with full force — it doesn't budge.
The Day of T0 Trial Mold: Eight Hours Staring at an Injection Molding Machine
The day the mold was cut, I took time off and went straight to the injection molding factory in Chang'an, Dongguan to oversee the T0 trial. From 8 PM to 4 AM, I stood in front of that injection molding machine. Not for show — the textured grain front cover was particularly difficult to dial in. The mold temperature had to be controlled at 85°C ± 2°C, the holding pressure time was trialed from 3.5 s to 6.2 s, and the shrinkage rate was adjusted from 0.5% to 0.7%. The operator changed the melt temperature five times and produced 49 scrap parts. The 50th one was finally acceptable.
The mold finishing master took one look and said "it's good," then set the part aside for natural cooling. In that moment, I finally understood what "making a surface" really means. When a mold master says "it's good," it's not a judgment — it's the muscle memory of hundreds of thousands of molded parts.
Looking back on this project, I think the biggest takeaway isn't some design methodology. It's confirmation of one thing: 80% of industrial design work doesn't happen in front of a computer — it happens on the production floor and in the user's real environment. You can render the most beautiful images in the world, but if a nurse can't use it comfortably or an injection molding line can't produce it, those renderings are just paper.
Alright, back to revising the drawings.
Originally published at https://www.hezidesign.com/jingyan/977.html
Originally published at https://www.hezidesign.com/jingyan/977.html
Top comments (0)