My Experience With a "Wearable" Medical Device

Everyone who likes my detailed heuristic gripes, I got a thing for you. For non-life-threatening reasons, I am wearing this for four days, and within 5 seconds I could tell that I was going to have a lot to say.

As usual, this is not to “be mean” or just to gripe for the sake of it. It is all a case study in how we can do better than this. Don’t fall into these traps with your products, whether hardware, software, or service, regardless of market.

Data Gathering

(This is on topic, bear with me) Modern complex machinery is computer controlled. Cars and trucks and backhoes and ships and combines and machining centers and printing presses and more are all gathering and processing lots of data all the time.

The devices gathering it use this data to operate properly themselves by tweaking their settings for power levels, alignment, pressure, fuel flow, and the surprisingly large number of instantly adjustable items that modern computer controlled machines have. They also trade some of that data with other devices. The engine talks to the transmission, for example, so instead of the transmission only handling what it is given, it gets advance notice, and details about what it should be getting. But otherwise, all this data is thrown away. Things like error logs are saved, but the 120 times a second raw data is just a stream that is used or not.

If you are trying to diagnose an issue, or are testing a product, you might want this data, so you attach a thing called a datalogger. These still exist as free-standing boxes, but are often just part of a piece of software on a tablet, phone, or computer.

They record some fraction of the data stream. Due mostly to old timey database constraints. But also storage space, and how much it matters. Once a second might be plenty. When testing, the datalogger will have a big red button on the side to capture events.

Events are things you, as the human, observe. You are driving and the car doesn’t shift right. You press the button, and that’s a point in time you (or an engineer more likely) can go back to, see what the systems were doing that caused it to not act right.

Then, the engineers decide if it is not in parameters, how to fix it based on the data stream. The test driver can often push other buttons, or has a co-driver with a clipboard, or a recorder, to say WHAT the problem was at the time they pressed the event button.

The Bravo Device

On Friday, I got a datalogger plugged into my body’s data bus. One of these: https://medtronic.com/covidien/en-us/products/reflux-testing/bravo-reflux-testing-system.html

I mean, I think. They never actually shared that level of detail with me, but I can read labels on the back of boxes and google stuff.

Of course, the connector to my body's databus is proprietary so they can't actually connect to it. So next best thing, the pill monitors directly... stuff. Again — and we're getting into the patient experience here — no one tells me exactly what it does. But we get the gist of it.

It is all about acid reflux-ish stuff, but the docs hate my symptoms so want to not just throw meds at me or do a possible-side-effects procedure, so are first gathering data. The pill is a data retriever. There is a box it connects to wirelessly which is the datalogger itself.

It retrieves data from the pill every 12 seconds. Nobody actually told me this, but the light blinks every 12 seconds and I know how engineers of such things think. I don’t know why in the video it takes so long to blink, but normally the blue light below the square button blinks every 12 seconds.

The unlabeled blue light. The one too bright at night, too dim during the day. Which is just baseline bad for this device. Let's look more…

Hardware

It's a rather thick and oddly shaped box, with buttons on what is functionally the end, so in normal wear you can only see them. As we’ll discuss later, you sorta have to see both faces and the angle doesn’t help unless you are doing all testing on a desk.

It is given to me in a hard partial case which makes it even thicker.

The case appears to be there almost entirely to cover up the power button and a mini USB port, which I bet is both charging and data synchronization when they get it back. It has no Wireless much less mobile connectivity, so the doctors aren't monitoring me in real time or anything.

This is like having an Expert Mode settings panel with a popup that says “please do not use.” If end users shouldn’t use these functions they should not exist. I’d probably have no power button, and have put contacts on the back (since I’d make it waterproof) and it docks into a cradle instead.

And the case should be rubber or fabric.

For completeness, here’s the manufacturer sticker if you want to look it up yourself.

Note that the company has been bought by Medtronic. There is no such thing as Given Imaging anymore. Anyway, the other thing the case has is a belt clip, which expressly should not be used as I’ll discuss later on, and some points to attach a neck strap.

This is really quite something. Since it's an entirely separate case they really had to go out of their way to make the case this bad. Note that it hangs at an entirely inappropriate angle. It is uncomfortable, hard to see, hard to use. The strap is also vastly too long. I'd say about twice as long as it should be. Yes I know how to adjust straps. At minimum length it’s on my abdomen, and I am not short.

By the time I left the hospital I decided this was completely stupid so what around the way home went and borrowed a chest pack from a friend.

E7t1WD-WEAEYD3e.jpg

The strap is stupid because:

  • Bad angle

  • Too long

  • Just a strap, so dangles and swings

  • The device is not waterproof

  • The case doesn't protect it very well

  • And most of all: It is a bit over $6500 and I had to sign a couple times to say I pay that out of pocket (no insurance will cover that) if lost, stolen, damaged, etc. So, let's talk about UX here. It has to be more or less on my chest at all times, 24x4 (4 days for this and most studies) but...

Is fragile, explicitly not waterproof, and is provided with a carrying system that almost guarantees it will swing around so get under faucets, in soups, crack children and pets in the head, etc. etc.

If that was the only problem that would be horrible, but there’s so much more.

Connectivity

Let’s first get back to the datalogger discussions, because I forgot to wrap back to that and it matters for the experience. The Event button exists here also. More on the many types, but you can’t just wear this, but have to have access to it all the time, to tell it when you do stuff or experience stuff. Heartburn? Push a button (though it's worse than that). Etc.

Apparently this is what the software works like when the data is downloaded.

Graphs of whatever it is monitoring, little vertical bars with flags for events the user entered.

Between the monitor pill thing and the box we’re looking at, it is wireless, thankfully. A wire, in this day and age of bad medical devices? No thanks. During the tedious onboarding presentation they say it's like WiFi (no, it's like BT) and has a range of 3 feet.

Oh, yeah. The onboarding is just terrible. There's a binder with a couple dozen pages of PowerPoint slides printed out they flip through. Naturally they read each one to you word for word.

There's no part of it where gives a full overview of the system, it's all just disclaimers, warnings, and what button to press. Oh, and jargon.

There's a very brief terrible and hard to read take home sheet, this is the only part that even sorta mentions the wireless connectivity.

Also note the terrible typography here; gray (which prints as dots!) and very thin type. For a medical device, so 100% of the population should be able to read it. I can barely read it!

So there's I guess the three feet, it can be arms reach away. It cannot. Bluetooth connectivity when sitting against my chest regularly, loses connectivity when it's a foot away.

You know it lost connectivity because they told you in the training and manual. Here it is having lost connection:

It screeches at you, and the blue light that blinks every 12 seconds doesn’t blink every 12 seconds. Sometimes the blue light doesn’t blink, sometimes it turns to a red light. That blinks every 12 seconds. Or… not? No actual lost-connectivity symbol, or onscreen instructions.

They think it's critically important and obvious that the [1] or [2] also disappear. I guess it has two available to channels to communicate on and that's what those are but it doesn't actually say that anywhere, and you’d think something else would make that clear, like an RSSI indicator (signal strength) and highlight of which one is active if any. people have phones, are used to BT, WiFi, and mobile network indicators. But how will almost any end users know what this actually means?

Oh, see that clock there in the corner? That's critically important, because that's the time that the data logger thinks the events occurred. Aside from 24 hour time that confuses many people in the US, it's an hour and a bit off of the real time locally. Yes, I picked up the recorder from a hospital unambiguously in my time zone.

The time matters because you have to keep a record of things you forgot to push the button this for or take other notes. They have to be logged at the device time or they're useless. Better: we wear it for days but there's no day shown on the device or on the log. As you can see later, the device knows what the date is, it just won’t tell us when we need to know it and that carried through to the paper chart, for some reason. I added it myself to the chart, sigh.

This is all obviously terrible design, software, systems, and experience. The time synchronization, presumably is some manual push a button to sync time thing, but why would you do that? Why doesn't the logger software automatically sync the clock on the device to the current time zone and DST offset?

I argue this sort of thing all the time on my projects, and sometimes I get my way for logical reasons, but a lot of times I do not get to improve it "because that's not the requirement."

Interaction

This thing, like a lot of stuff we use everyday, clearly had an industrial designer take a stab at making the hardware swoopy and picking the colors, and otherwise just engineers built the whole thing.

Aside from logging what is happening in my esophagus, I'm supposed to press buttons tell the logger what I feel and what I'm doing. Here's is the interactive side:

Clear as day!

Here's the tiny thin grey text guide for it. First, the onboarding Powerpoint-like guide has the same chart, but… upside down. I have the device in hand, and takes me a good bit of time to realize they started on the right. So bad, and inconsistent between the two guides.

Anyway, horrible icons. Not the shapes, the specific ones. The left one is not food (knife and fork) but is any thing you ingest that's not water. Drinks also. And yes, why shapes? Why just shapes!?!?!

The screen is supposed to help you I guess. Noticed the little debossed triangles in the plastic between the buttons and screen? The center three are supposed to be softkeys — correspond to the on screen display — but thats not the right way to do it. Also, pointless. Now each function has two different icons, no words, and they aren't great icons either.

Also, left and right arrows? Carriage return? Escape? I assume those only work when you get to expert mode, which I should not. But I’d definitely not let that happen so have decluttered the interface to not have those symbols shown to end users.

It gets better and better. For starters you can't actually press one of these buttons to log an event.

The buttons only log the event when the screen is on. Turn the screen on by pressing a button. Faint beep says it worked. Light on key means You Pressed The Key!!! Yes, even if the only action was turning on the screen!!!

The two outboard buttons are events with a start and stop time. When you start eating or drinking something you press the left button, probably twice, to go into eating and drinking mode.

Remember, this button press isn't reflected on the screen, at all, so the only way to tell if worked and you're in that mode is that the button light blinks. Then you have to remember to unlock the screen and press it again to turn it off.

And then look at it for a while to tell if that worked. Because remember the button press lights up the button light. So now you are looking for /absence/ of blinking. Blinking is effing terrible. Always was, but in LED era, must be banned: https://uxmatters.com/mt/archives/2019/05/a-design-language-for-touch-gesture-and-motion.php…

The buttons are not usefully tactile. They are membranes with no fences, and only a slight embossing across the whole surface so you can’t really find them, and cannot tell them apart. Embossed shapes might have made the three big shapes make some sense, as circle/square/triangle are differentiable by touch. But hno.

And the buttons do not have backlight when you do manage to find one to turn the screen on. Again to the not-waterproof etc part: Because everyone will use this in full daylight sitting at a desk.

Beeps might help to confirm actions, but for the part they are quiet beeps. How about haptics to be detectable in all environments (even just people talking can drown out the beep)? How about different beep/vibrate for different actions? Nope, none of that!

One time I looked, and it’s been in heartburn “mode” I guess, for a while. Screen is unlocked long after the last interaction, and this on screen symbol is highlighted, I guess.

I got out with press and hold on the same button, so… I guess it has secret functions also. Simple interfaces shouldn’t do secret things. The staff had never heard of it doing this, so I mean it when I say secret. Doesn’t even come up in internal training.

The button LEDs are green. The unlabeled vague maybe-connected LED is blue. At a glance (esp when not in foveal vision range where you cannot see color well or at all) the maybe-connected light can look like you left eating or sleeping mode on by accident. Also, all are the same shape. LEDs do not have to be tiny squares or dots, but can be any shape; they could have chosen to light up the whole button.

That's when you can see them. If you don’t position your finger carefully, you obscure it, cannot tell it was clicked. Outside? No. No chance. Sunlight entirely blows out the LEDs so they cannot be seen. So, am I in eating mode or not? Can't tell. And the mode switch is a pressed toggle so I can't turn it positively OFF.

To the observed events, the three center buttons, what all do those mean again? Well, a lot of things.

I was told to use the circle button not just to me that I threw up a little, but for anything that comes out of my head like burping or coughing. The maker has this paper log also.

Which aside from conflicting with the instructions I got, you're apparently supposed to keep with you at all times, with a pen, and write down all kinds of other events. A letter sized piece of paper, and a pen. And, look at the box top get the time.

I write down the meds, and a handful of other notes, but otherwise that's clearly insane and stupid. You're going to end up with selection bias issues so I'm not writing down anything and using only the buttons whenever I possibly can.

Don’t Design For the Wrong User

At the end of the “study” — that’s what they call the sampling period for this — the recording device has to be returned so they can plug it in and get the data. Much to my surprise, it doesn’t just keep recording till then, but it records for Ninety Six Hours, and then stops.

Then, it tells you…

Well, it depends on who “you” are. Because it is clear now that it is not designed for actual patients to wear and push buttons. This message will, 100% of the time, come up when patients still have it, and are maybe on the way to the hospital to give it back.

So, it says something to the tech who gets it, later. There are steps. Assuming the entire experience happens on the desk, and there is no environment, no users.

Does it even just say “done”? Nope, it is the worst of the engineering mindset, and defaults to telling you what you did wrong. It is not success, but an error message.

Don’t be that way.

Shameless Plug

I do this sort of stuff formally, for work, all the time. Heuristic evaluations, or expert reviews. Usually less snarkily, and in a spreadsheet. If you want me to gripe about your product in even more detail, I'll do that if you ask.

If you also pay me, I'll only send the report to you even, so you can fix it instead of being shamed in public.





BloggingSteven Hoober