Accurate Prediction: The Next Phase in Diabetes Innovation

A few weeks ago, we talked about prediction modelling and its use in diabetes research. In this post, we look at the power of prediction and how it can enhance the technology used by the artificial pancreas.

 

What is the artificial pancreas?

Artificial Pancreas

Image via Mirror

Back in 2015, a 4-year-old boy named Xavier Hames (pictured above) became the first Type 1 Diabetic to receive an artificial pancreas. The artificial pancreas (AP) automates blood glucose management and reduces the likelihood of hypoglycemia or hyperglycemia. Much like an actual pancreas, the AP provides insulin when needed.

 

How does the artificial pancreas work?

The artificial pancreas has 3 parts to it, as illustrated below.

Artificial Pancreas vs Accurate Predictions

Image via FDA

The first part is a CGM sensor that sits under your skin and measures the blood glucose levels in your interstitial fluid. A transmitter, which sits on top of the sensor, sends this data to a receiver. CGM data needs to be calibrated with a blood glucose meter.

The second part is a computer-controlled algorithm, which receives the information from the CGM, and performs mathematical calculations. Based on these calculations, the controller sends instructions to the insulin pump on how much insulin to produce. The insulin pump (the third part) adjusts accordingly. Presently, the AP system only releases insulin. However, pending FDA approval, it will release glucagon as well.

The fourth “part” is your body. Since glucose levels are constantly affected by diet and exercise, you play a critical role in making important adjustments to lead a healthy life.

Since the AP would rely on a CGM, a predictive algorithm that accounts for the delays associated with the CGM would be vital to the success of the technology. Furthermore, the AP technology needs to distinguish between different situations. For example, it would need to distinguish between a user who is napping and a user who is running to catch a bus, with little or no input from the user. The AP needs to distinguish between these events and connect to the user and their environment, in order to administer the right amount of insulin.

 

What is an accurate prediction?

Accurate Prediction

Foreseeing this limitation, we developed Docto, an app that uses machine learning to predict your blood glucose levels one hour into the future. Using a CGM and fitness tracker, Docto will learn trends in your blood glucose levels, and predict future outcomes.

We believe that our technology will help make diabetes management considerably easier. While knowledge of the past is important, lack of foresight can leave Type 1s in a perpetual state of uncertainty. And we believe that you should never have to feel uncertain when it comes to your own health. Moreover, we believe that you need to have a good idea of what state you’re in and what state you’ll be in, to be in complete control of your diabetes management.

Thus, we believe our technology allows you to see the best of both worlds: your past data and future data. The more that you use Docto, the more it will learn about you and from you. And, like a human, it will adjust accordingly, learn from its mistakes, and be able to give you an accurate prediction of your future. For more on Docto’s accuracy, click here

***

Interested in learning more about how Docto works? Check out this post. Additionally, make sure to follow us on Facebook and Twitter for more articles about Type 1 Diabetes innovation and research.

Download Button


Comments (0)

Leave a comment