Developing the technology which could predict blood glucose levels and prevent serious complications for people with Diabetes is the ultimate goal for Afon Technology’s artificial intelligence (AI) experts.
We spoke to Afon’s Brad Love and Dan Fowles to take a closer look at how the company is harnessing big data and is paving the way for personalised health solutions for people with Diabetes.
Transforming science through data
When Brad Love mistakenly sat a maths entrance exam as part of his application to study law at university, he had no idea just where it would lead him.
Brad, who is now a Statistician at Afon, ended up passing the maths exam with flying colours, leading him into a career working with statistics and big data in the field of biotech.
Brad and his colleague Dan Fowles, who has a PhD in AI and is an AI Data Analyst at Afon, have together developed the software, algorithms and data analysis that underpin much of what Afon does.
Afon is working on a ground-breaking needle-free continuous blood glucose monitor, Glucowear™, which uses low power RF/microwave technology to track and record changes in blood glucose levels, in real time.
The data gathered by the device has the potential to transform how people manage their condition, Dan says.
“Typically, an individual’s glucose levels will change over time,” he said. “There’ll be some kind of pattern based on their normal routines throughout the day, or over weeks or months, and so a pattern will start to emerge in this individual or over a group of individuals. What we can do with the AI is we can learn those trends or use it to help with our predictions.
“By creating a device that is real time, it provides you with almost instantaneously available glucose values. There is no delay, which may cause problems or risks associated with lower high glucose values.
“When your blood glucose suddenly starts dramatically dropping, that generally means that you are a short time away from passing out or losing consciousness. And if you can intervene, you can prevent this medical issue. If your blood glucose is shooting up rapidly, you want to start taking some insulin right away to knock that down.
“The more that your blood glucose starts bouncing back and forth, it starts to accumulate damage inside your body. So, the real time aspect really allows you to start taking action on what your body is doing and to help you blunt that effect on your system.”
Brad said one of the biggest challenges in AI is accommodating the differences between people, saying: “There is a tremendous difference between how people’s glucose behaves from person to person – we’re starting to learn this. Not even just blood glucose, but how other psychological parameters react.”
What happens when big data “flexes its muscles”
While Brad and Dan both say that AI is only as good as the information you provide, its potential to provide both an overview of trends alongside personalised patterns could benefit everyone with Diabetes.
Brad explained: “AI is continually collecting data, and as you ask it more questions, it will help provide you these answers. And this is where big data really starts to flex its muscle.
“We can actually start to have the app ask you questions like, what did you just eat? And we can start to learn and then as we accumulate this for more and more folks, we can crowd surf information and see how people’s bodies are reacting and pass out that knowledge to everyone.
“The idea of what can be done is tantalising to me. I think that’s the vision of personalised medical tracking that we start to see.”
Both Brad and Dan are excited about the prospect of how data and AI could be utilised to predict blood glucose levels, enabling people to take preventative action. This idea of using past blood glucose readings to predict future levels and provide bespoke health solutions is the “long-term vision”, they say.
Brad said: “I think that is the good long-term goal. Good AI and good algorithms and good tools in general allow you the ability to build information and build knowledge.
“Using it to tell us what’s happening right now with your blood glucose is great. Turning it around and letting us know what’s going to happen 15 minutes from now, 20 minutes from now, 30 minutes from now, so that you can do something to prevent a bad situation – that’s a goal.”
Despite the challenges of developing and testing AI, Brad said the satisfaction lies in getting it right, saying: “When it all works, and you say this is what the blood glucose is, and then somebody looks at the blood results and you’re like, wow, there it is – that’s the beauty.
“And to know in the background how many things had to go right. It’s not just about software, it’s not just about statistics, not about hardware – knowing that the dedication of the whole team made this magic happen, is really quite incredible.”
