Connected pens

Data scientists are entering promising, uncharted territory

Through novel data-driven insights into patients’ real-life behaviour, data scientists can finally illuminate a century-old problem.

Data science is now – finally, one might add – a vital piece of the puzzle in helping the more than 530,000,000 people living with diabetes worldwide.

By analysing data on patients’ real-life behaviour, collected through a so-called ‘smart pen digital ecosystem’, data scientists in Novo Nordisk might redefine how we understand and support diabetes treatment.

The ‘smart pen digital ecosystem’ is simple: A smart pen for injecting insulin + a continuous blood glucose monitor (CGM) worn on a patient's body. Both are connected to an app that gives patients an overview of their numbers, glucose and injection data, medicine dosage, and the like.

Let’s break down the components

The pen first, starting with a bit of history. Since insulin was discovered as a treatment for diabetes in 1922, there has been a need for a simple way to inject the medicine. One of the first products to help do so was the Novo Syringe, invented by the founders of Novo Nordisk in 1925. While definitely not solving all problems, the idea of a product that patients themselves could use to inject their medicine lived on and later gave birth to insulin pens. The first Novo Nordisk Insulin pen was launched in 1985 and made injecting insulin a whole lot easier.

However, patients have still been dealing with the problem of recording and logging the injection data. When did I take the last dosage? What was the dosage? Did I even take one? Until recently, patients themselves had to manually log down their injection data, leaving room for well-known human error in this vital data collection process.

The Novo Syringe from 1922

Enter the second component: data

Insulin pens have come a long way since 1985 and have now become smart. The smart pens automatically record and store dosing information about every injection. This data can then be transferred to an app which is linked with a CGM. Today, there are several CGM devices which patients can use, but they all share the core function of continuously providing patients with estimates of their blood glucose level. Combined with the smart pen, patients are able to monitor their glucose levels alongside their injection data.

Essentially, the smart pen, in connection with an app, remembers everything important – so patients don’t have to. Patients can then share this data with their doctors to make more data-driven and personalised decisions on their treatment.

The NovoPen® from 1985

Smart

Essentially, the smart pen, in connection with an app, remembers everything important – so patients don’t have to. Patients can then share this data with their doctors to make more data-driven and personalised.

The data scientists enter the picture 

So, does it work? In the first Novo Nordisk smart pen pilot project from 2020 on 94 patients in Sweden, the answer is yes, it does:

“We analysed the data and evaluated if patients were in better control of their disease after they initiated the smart pen in combination with CGM monitoring. We found that the patients in the pilot study improved their glucose control and adherence, so they missed fewer insulin injections.”
Nikoline Nygård Knudsen, Global Data Strategy Lead at Novo Nordisk (Reference: “Increased Time in Range and Fewer Missed Bolus Injections After Introduction of a Smart Connected Insulin Pen”: Adolfsson et al., 2020.)

A small step for 94 people, a big step for the more than 500 million people living with diabetes. The data quest has just begun.

Today, smart pens enable patient data on glucose levels and insulin dosing to be collected via our partners' applications and then transferred to the Novo Nordisk data lake (all data is anonymised and in GDPR compliance). And it’s a lot.

We get a glucose reading every 3-5 minutes. That’s over a billion data points by now (1.280.000.000, to be exact). More than 60,000 smart pens are in patients' hands, and more than 20 million injections have been uploaded. And the data keeps flowing.

The most important job of the data scientists working with projects like these in Novo Nordisk is to analyse this data – to better understand our patients’ behaviours and make it easier for them to manage their disease.

"The data are interesting because it is real-world data on a large scale, capturing the everyday life of patients with diabetes. Thus, there are many interesting stories in it to explore and curate. It is an excellent example of how you can speak with data and change diabetes. To do so, we use technologies such as Amazon’s Redshift databases, Databricks, Posit Workbench and Posit Connect, R, Python, R Shiny dashboards and Streamlit. Azure DevOps, Git, TensorFlow."
Niels Væver Hartvig, Principal Data Scientist at Novo Nordisk

It’s a complicated matter. It’s an entirely novel data set that requires an in-depth understanding of patients' real-life behaviour to make sense of the data patterns and irregularities. It truly is uncharted territory. Being so, our data scientists have a special obligation to share their findings with the rest of the world:

“Novo Nordisk is the first player in the field to have this type of data. We have produced insulin for 100 years, and never before have we known how people have taken their insulin. Now, we have data that shows exactly what the patients do. Now, the patients can finally see their data. They can share it with their doctors to improve their treatment. That’s the promise of the data. We have a scientific obligation to go out with these results.”
Nikoline Nygård Knudsen, Global Data Strategy Lead at Novo Nordisk

True knowledge-sharing

In 2023, Novo Nordisk is, together with key doctors in the field of diabetes, presenting five new abstracts at different scientific conferences. Our data scientists are co-authors on all of them.

Working as a data scientist in life science is, in that sense, very different from other industries. Here, data scientists are part of a cross-organisational group discussing what research areas should be investigated and how.

Working with medical researchers, data engineers, data architects, software developers and many other talented people to develop projects like these and share them with the scientific community is not only an academic endeavour. It’s through processes like these that you can see your work make a difference:

“I think most people are driven by making a difference for real people out there. Here, you don’t feel so far away from that. We talk to doctors and physicians at conferences, and we hear upfront how the smart pens and the digital ecosystem creates a new foundation for the dialogue between them and their patients – and it’s the exact same data we are looking at and trying to understand.”
Nikoline Nygård Knudsen, Global Data Strategy Lead at Novo Nordisk

Today, data scientists can finally use reliable data to shed light on a century-old problem in life science. Can the data live up to its promise? That’s up to ingenious data scientists to find out. Someone like Nicki here. Or maybe someone like you?

“On one of my first days here, I read a project description where part of the project's principles was described as something along the lines of 'prioritising people's mental well-being above meeting project deadlines.' I had never seen anything like that. Novo Nordisk goes the extra mile in implementing our values in this area.”
Nicki Mørk Bolbroe, Product Director in Data Science at Novo Nordisk

Didn’t find what you’re looking for? Right now, there are more than +1000 job openings at Novo Nordisk. We’re hoping one of those has your name on it.