AI

Will an artificial pancreas be a gamechanger for
type 1 diabetes?

In December 2019, French start-up Diabeloop announced they had secured €31m Series B funding (a record for a European therapeutic AI company). Its device, which has been described as an 'artificial pancreas', uses artificial intelligence to automate and personalise treatment for patients with type 1 diabetes. Abi Millar finds out more. 

F

or patients with type 1 diabetes, managing their condition can be an ongoing battle. Patients need regular blood glucose monitoring, daily insulin treatment and constant vigilance, on top of maintaining an otherwise healthy lifestyle.


The condition usually starts in childhood and makes up around 5%-10% of all diabetes cases, with a global prevalence of up to 20 million. An autoimmune disease, it occurs when the immune system attacks beta cells in the pancreas, meaning the body can no longer produce enough insulin. Treatment focuses on keeping blood glucose around the normal range, avoiding the dangerous extremes of hypo- and hyperglycaemia.


For many type 1 diabetes patients, their treatment regime can seriously curb their quality of life. Because the condition has to be monitored so precisely, it can mean sacrificing spontaneity and a sense of freedom.


The good news is that diabetes management has improved dramatically in recent years. In the past, patients needed to perform regular finger-prick tests with a portable glucometer. Nowadays, many patients use a continuous glucose monitoring (CGM) sensor, which enables them to check their blood sugar at any time. Sometimes, this is used in conjunction with an insulin pump, which delivers tiny amounts of insulin into the blood 24/7.


While these devices are not yet widely used, we are rapidly moving towards a new era, in which technologies play the role of an artificial pancreas.


“Our belief is that, within five to ten years, there will be a limited number of patients who won’t have a system that is fully automated,” says co-CEO of Diabeloop Marc Julien. “While there are always some patients who don’t want to have anything on their body, we believe around 70%-80% of all patients will have an intelligent device that manages their condition for them.”

Automating diabetes management

Diabeloop, a Grenoble-based start-up, is one of the companies at the forefront of this new wave of diabetes management. In December, the company announced it had won €31m in series B funding (breaking the record for a European therapeutic AI company). The funds will be used to bring their device to market, first in Europe and eventually in the US.


“Our device has a CGM and an insulin pump, which are currently the standard of care on the market,” says Julien. “What we add is a dedicated handset that goes in between, connected by Bluetooth low energy. Every five minutes, the system calculates how much insulin you need and gives the order directly to the pump. So the patient is free from having to make these decisions.”

“The device is fully automated, and all the patient has to do is carry it around.”

With a CGM taken alone, the patient has to check their blood sugar levels and then make a call on whether or not they need insulin. In this case, a self-learning algorithm does the work on their behalf. It analyses the data in real time, assessing the patient’s physiology and other factors to determine how much insulin to administer.


This means the device is fully automated, and all the patient has to do is carry it around. It should be noted there’s one exception – every time the person eats a meal or does some exercise, they need to manually input that information.


“Some people don’t provide that information, but that’s not a good idea because we are better at detecting the future if you do,” says Julien. “Your body will have a different physiology and reaction to the insulin than if you hadn’t exercised or eaten.”

The roadmap to approval

Julien founded the company in 2015, together with his chief medical officer Dr Guillaume Charpentier, a diabetologist, and co-CEO Erik Huneker, an engineer.


“Some of Dr Charpentier’s patients were living very far from his hospital, so he wanted to give them some technology that would work like a doctor in their pocket,” says Julien. “He developed a couple of CGM tools, but thought why not put intelligence in it? So he developed an academic proof of concept, and that’s when Erik Huneker looked at the project and said we should turn it into a product.”


Over the next few years, the device was redeveloped and underwent extensive clinical testing. It eventually received its CE marking in 2018.


“Following EU approval, we started to engage with payers in France and in Germany to get their assessment of the efficiency of the system, and see whether they could pay for it,” says Julien. “We’re looking to market this solution in Europe as soon as possible, and in parallel we’re in discussions with the FDA to gain approval in the US. We’re launching a major clinical trial to provide the data the FDA needs.”

“We’re launching a major clinical trial to provide the data the FDA needs”

Over the past few years, Diabeloop has benefited from an IPME Phase 2 grant by Horizon 2020 and support from EIT Health. The latest funding round came from a broad syndicate of investors, some new and some historical.


“We have three types of backers. One is specific venture funds dedicated to our kind of industry, some of which have been with us since the inception,” says Julien. “We also have some investor partners that have been helping us develop, and a third category is entrepreneurs who believe we have a breakthrough solution for patients and want to help us get on the market.”


At present, the company is investing in tweaks to the algorithm, to make it suitable for different types of pathology.


“For example, this system is adapted for adults but we have also developed a specific solution for kids that will need to be promoted and pushed on the market,” says Julien. “We also want to help patients who have very unstable diabetes that is difficult to control, so we have a new solution in clinical trials.”

The impact for patients

The device, then, could be a game changer from a technological standpoint. However, the really critical factor is the difference it makes to patients’ lives.


“We have had amazing feedback,” says Julien. “Doctors can see how much better-controlled the patients are, and the patients feel that they’re having fewer episodes of hypo- or hyperglycaemia. But more importantly, they’re now allowing themselves to do things they couldn’t do before. We had feedback from one patient who said to us, for the first time since I was diagnosed I went to walk alone in a forest.”


There are also implications for the patient’s family, who may have been integrally involved with their care.

“One patient said] for the first time since I was diagnosed I went to walk alone in a forest.”

“Another person said, this is the first weekend we all slept through the night,” says Julien. “If you’re a mother looking after a kid, you’ll be waking up repeatedly to check the kid is OK, and it’s the same if your partner is next to you. This gives you the ability to free your mind from these worries.”


He hopes that within the next few years, all the patients who need this kind of system will be able to reap the benefits.


“The relief, the ease of use, the change in their lives, is just so massive,” he says. “There are two million patients in Europe and 1.5 million in the US, so that’s the potential market for this device, depending on how much it’s pushed by doctors and who’s paying.”