As a passion project, I gave myself 5 days to research and rethink an insulin pump OS and design a remote control app based on first-hand feedback. This insulin pump and app allow diabetics to initiate insulin injections through a UI that also receives and visualizes glucose data from external devices. I was mainly interested in coming up with an improved and intuitive flow to complete the most common actions and rethink the UI and interactions to strengthen the relationship between device and user.
Arguably, the most crucial, fundamental and frequent action of an insulin pump is to bolus (initiate insulin). The process is fairly lengthy, requires a lot of user effort to be successful.
Diabetes is one of the most common autoimmune disorders. I have been a diabetic since I was a child and ever since I was intrigued by how technology could make handling such a widespread disease easier and more enjoyable. I know from personal experience that my diabetic fitness was in better shape if I was engaged and in love with the devices and supplies I used.
Within 5 days I did light weight research, interviewed 80 diabetics using pumps and ultimately designed a modern approach for a diabetic pump OS. This product is technically a wearable, a completely new space for me. I was excited.
It takes too many steps to complete a bolus – the most important action.
Reduce input and compress views. By combining information into a single view and using available data vs having the user input it can save time and effort.
Interactions and IA are dated, confusing and too nested.
Simplify IA and adapt modern interactions. Instead of clicking scroll arrows, scroll. Instead of tapping 3 times, use a single gesture. Instead of nesting screens multiple times, combine data in single views.
Dated, uninspiring and inaccessible look and feel.
Modernize UI and enhance accessibility. Meet modern design standards and make it delightful. More accessibility, higher contrast colors, less chrome, more content.
The beginning and result of this challenge. Since the device is black, the first choice I made was to also color the background black to allow for a seamless transition. Simplified controls, fewer elements on the screen and emphasized touch targets for common actions.
Original OS design on the left vs proposed design on the right.
This view gives users an overview of their glucose trend, how much insulin is currently active and device statuses such as battery and insulin level. The majority action will be to initiate an insulin bolus from here.
The pump is being unlocked about 40 times every single day. Locking the device is a crucial safety measure to prevent accidental insulin boluses. A simple swipe will get you past the lock screen.
The current method on existing devices requires the user to tap 3 different areas of the screen to unlock the device. It requires a lot of effort, this effect is being compounded by the fact that the process needs to be restarted every time the user mis-taps.
I emphasized the bolus button. This button will be hit more than any other button on this pump. Tapping the Graph button will change the time scale by a set increment.
The pumps communicates with other devices such as glucose monitoring systems. The data will be passed to the pump for bolus calculations and general observation of sugar trends.
Dragging over the screen moves the time scale ticker and shows the value at each point in time.
This is the most frequently done action on the pump. The existing device requires over 6 taps in the best case (no adjustments necessary). I have tried simplifying the process as well as populate data automatically instead of relying on the user to do this.
Step 1. Hit the Bolus button.
Step 2. Enter the amount of carbs you want to eat.
Step 3. Review your choices and hold the bolus button for one full second to deliver insulin. This eliminates another confirmation screen.
Success. That's it for the basic 80% bolus case. We reduced the amount of necessary actions to a minimum, but still offer options should a user want to fine tune their bolus.
Based on personal experience as well as my research I have learned that the pump community desperately wanted an app on their phones to bolus from. This would add convenience, extra features and most importantly, more privacy. No need to pull out a pump from your belt/purse and fiddle while out and about.
The app should feel like natural extension of the pump.
Among the millions of diabetics are children and infants. During my interviews with real users, parental controls were among the top asks. Being able to see how much insulin was already used and who initiated it could prevent mishaps and potentially life-endangering situations.
The app should feel like natural extension of the pump.
I knew that changing the a few user flows and redesigning the main parts of the OS wouldn't solve the problem. There are so many more considerations like the IA, user and device settings, relationships between screens as well as views that are not being displayed frequently, but should still be accounted for. I considered most screens to make sure this new system worked holistically.
When I first thought about getting an insulin pump for myself, I did the usual research online – what does the product do/not do, what is the 5-year roadmap, what are common issues users are having? What are the benefits, what are the cons? Pumps have been around for decades, but only recently starting becoming more user-friendly. Even before I received my actual unit, I thought about ways to improve it.
However, me being brand new to this space, I wasn't the best person to ask for feedback or suggestions when it comes to improvements.
I joined a highly active Facebook group for Tandem pump users (the brand of my device). On day 1 of this challenge, I asked the community what things they missed about their pump and what they think could be better/easier. I have received over 200 comments and personal messages.
• The lack of remotely giving insulin was a major frustration
• Unlocking the device was cumbersome
• Amount of confirmation screens
• Amount of audible notifications and alarms are tedious
• The pump requires too many manual inputs when it could easily use existing data (using a bluetooth glucose meter)
• Seeing remaining insulin from a phone would be great
• Bad quality device clip and case
• Device home button not responsive enough
• Insulin not visible in cartridge
• Beeping during insulin fill is annoying
• Cartridge filling process
On day 5 of this challenge, I made another post on the Facebook group to show everyone what I came up with based on their previous feedback. I collected the various feedback and curated a list of priorities for the next round of iterations. Overall, the overhaul was positively perceived and the community seemed engaged.
"I love all the info is on the app! I get so tired of asking my 13 year old that!"
"Love this idea. It would make it so much easier."
"Parental controls are amazing. This way I can check my wife dosed vs when I try to dose. Say we are at a party"
"Amazing work! I would love to be able to bolus for my son via phone!"
"This makes me want to use my pump all the time"
"I like that the app and the pump match in look"
"Great work, I really like the hold to deliver idea for a bolus too!"
"Something I would like on the pump display would be hour indicators. Perhaps lines, like in upper right of last pic, for hours and similar dashed for half hours. As an option."
"I’d love to see my IOB on the BG graph!"
"Is there a widget with this app ? Would be nice to see parts without unlocking the phone."
"Amazing work! I would love to be able to bolus for my son via phone!"
"This makes me want to use my pump all the time"
Working on this was not only a ton of fun, but also eye opening for me. After talking to the community, I learned about things I haven't even heard of before. Since I was so new to this product to begin with, it felt like I was not ready to tackle this problem as I don't know nearly enough about it.
In hindsight, I was glad I did not give up on it. I have a much better understanding not only of the product itself, but also its restrictions and regulations in the medical field.
If I had more time, I would've liked to test my designs with users. Run sprints with them and come up with a breadth of various options to see which one makes most sense for each scenario.
These designs have been handed off to Tandem for their consideration.