I wanted to spend some time expanding on this post and share more of the work we’re doing to make diabetic data more approachable and useful to both patients and caregivers.

In the ecosystem, there’s an ocean of data – it’s not very easy to aggregate not consume. Once you solve the hurdle of aggregation, how do you utilize all that information? And where do you start?

Data visualization itself can be considered a science and art but it shouldn’t take an advanced degree to understand information that’s supposed to keep you alive and healthy.

Let’s take a look at this snapshot of a typical Nightscout dashboard (for one single user):

If you’re a diabetic patient familiar with your CGM data this may look familiar or unfamiliar.

Let’s break down the view we have.

Focusing on the left corner, we have:

• The current time
• When the last blood glucose reading occurred
• Insulin pump status: battery level, last bolus, basal rate, basal profile, time left of the current basal rate, etc…(accessible by hovering over the tile as shown here)
• OpenAPS status: carbs on board, insulin sensitivity factor, target blood sugar, prediction values, insulin sensitivity ratios, etc… (accessible via hovering over the tile)

In the center, we have a three hour graph of a patients blood glucose indicated by the gray scatter plot.

The purple scatter polot indicates predictive blood sugar levels going into the future beyond 10AM.

Moving to the top right:

• The user’s current blood sugar value from their CGM
• Insulin on board, carbohydrates on board
• Change of blood sugar since the last reading
• Insulin pump cannula age, insulin sage, sensor age

At the very bottom, we have a miniature graph and timeline showing a scatter plot from the prior day.

That is A LOT of information in one place, isn’t it? And more questions surface:

• What’s actually important?
• What do I do with all this information?

Let’s take a small detour… When I was leading the on-call team for Netflix’s website, we had the responsibility of handling incidents and outages as we scaled globally.

During that time, we learned a lot around building dashboards and displaying operational information that’s important to triaging or understanding a system.

One of the most important lessons: Present only the information that’s important.

But what’s important in the diabetic context?

• The current blood glucose
• Some historical data (prior 3 – 4 hours)
• Some future prediction data (at most 3 – 4 future hours)
• High level device status (online, battery percentage, etc…)

After talking with lots of users and conducting interviews – most of the Nightscout data is useful but it’s not presented in a user-friendly manner.

How does Serendipity do this differently?

Serendipity’s main dashboard removes and reorganizes the information into hierarchies. The main chart provides you with a 3 hour window of your CGM data with the separate graph below illustrating the prior day’s CGM data.

High level device information is displayed to the left without the user needing to hover over small tiles.

The goal of the Serendipity dashboard is to present only relevant information to the user so they can take action (or do nothing). At a glance the patient should know whether their blood sugar is within range or headed high or low.

Below is a snapshot of the same patient’s data in the Serendipity platform:

The keen observer may notice that there’s a lot of information “missing”. That’s intentional. The dashboard has started with the bare minimum to help reduce the noise and energy it takes to understand where the patient’s glucose is headed. Oh, and I forgot to mention – this is all mobile friendly.

This is still a work in progress as we’re still working on what makes sense to display for the different devices in this main view versus things that get organized to other parts of the platform.

The goal here is to surface information that’s immediately actionable to the user while other pieces of information should be shared in other parts of the UI or even notifications pushed to the user’s mobile devices, email, etc…

I’d love to hear back from any of you that have thoughts, ideas, or even additional use cases we should be looking into – this includes Type-1 and Type-2 diabetics, physicians, nurses, or data visualization folks.

Don’t hesitate to drop me an email!

Have someone that would benefit from using Serendipity? Sign up here.

All the best,

Jonathan