Professor Roman Hovorka created the world's most advanced control system for the artificial pancreas. He explains how close we are to an autonomous insulin system for type 1 and type 2 diabetes, and the barriers to overcome.
Forward: features are independent pieces written for Mewburn Ellis discussing and celebrating the best of innovation and exploration from the scientific and entrepreneurial worlds.
“There is a report saying people with type 1 diabetes think about it every eight minutes,” says Professor Roman Hovorka. “It is gruelling.”
Professor Hovorka has spent the last three decades trying to change this. He's the lead at the Institute of Metabolic Science at the University of Cambridge on the project to optimise insulin pump control systems so, ideally, the user can forget about their condition. His work holds the promise of ending the tyranny of diabetes.
Professor Roman Hovorka, Founder of CamDiab
The challenge of building a fully-autonomous insulin mechanism is immense. Blood glucose sensors and insulin pumps operate with precision, but programming them to inject the correct amount of insulin into the body is fiendishly difficult. Pumps deliver a continuous low dose of insulin, known as the baseline or basal rate. In closed-loop systems, this basal delivery may be automatically adjusted in response to glucose measurements. However, when a boost is required, for example at mealtimes, a larger insulin dose known as a bolus is delivered, which, in a hybrid closed-loop (HCL) system, must be initiated manually.
Professor Hovorka is the world's leading authority on the closed-loop control algorithm. He developed his first algorithm in 1997, and today continues to refine his work. He is the founder of Cambridge spin-out CamDiab, which makes the CamAPS FX smartphone app for controlling glucose levels in type 1 diabetics. The app runs on Android and iPhone, and works with continuous glucose monitors made by Dexcom and FreeStyle, two of the major global brands, and with YpsoPump and Dana insulin pumps.
“We have about 100,000 users in 18 countries,” he reveals. “We have a pregnancy specific closed-loop system. We had an exciting report by NHS England which showed that 80 to 90% of type 1 diabetic women in England are using CamAPS FX, and we achieved the best ever pregnancy outcomes in the region. Both for maternal and neonatal. It's fantastic to hear.”
The search for perfection
The current HCL setup run by the CamAPS FX system is already hugely impressive: life-changing for people living with type 1 diabetes. Beforehand, a person with diabetes was required to administer insulin manually using injection pens two to five times a day, in extreme cases up to 10 times.
Roger, who made the switch to HCL in 2019, said:
“Now when we go out for meals all I need do is take out my smartphone, enter the carbohydrate count and away it goes. I can see my levels instantly and can do everything I need to do on the phone. Before I had the app I needed to carry around with me needles and insulin and would go to the toilets to do a finger prick test and injections as needed. That’s all taken care of now with complete discretion.”
The phone app means his wife can check his levels remotely. When Roger fell asleep she worried he was going into a diabetic coma: the app showed his blood glucose levels were within range.
The mental burden for users is therefore dramatically reduced.
Moreover, with the recent release of CamAPS Liberty feature, users will soon have the option for fully-autonomous insulin management, known as a fully closed-loop (FCL) or artificial pancreas, that will enable users to temporarily pause the requirement to input carbohydrates at mealtimes, depending on their personal needs.
Since the HCL algorithm relies on user input, user adherence affects algorithmic performance – if the user does not enter meals/exercise, the system has to chase the resulting glucose level rises/falls. The FCL system aims to eliminate user input, by replacing user carbohydrate counting and pre-meal bolusing with real‑time predictive control. However, without meal/exercise announcements, the FCL algorithm must infer meals/exercise from CGM signals, i.e. when glucose levels are already rising/falling. This makes FCL system design challenging – the lack of advanced warning and responding when the meals/exercise are already showing in glucose levels.
So how close are we to reaching perfection? “It's a good question!” says Professor Hovorka. “There are better components arriving frequently. There are new blood-glucose sensors and new pumps, which are delivering improvements. There's a new sensor from Abbott which measures ketones. If someone has high ketones it means they aren't getting enough insulin and something is failing. So the technology outside the algorithm continues to get better.”
The control algorithm performs well, but the leap from good to perfect is a big one. “We face real life!” he admits. “There is a mum with two or three kids, handling things, and she doesn't have the time to do everything we ask of her.” Furthermore, users forget to refill the pump insulin reservoir. They eat unsuitable things. And they exercise in unpredictable ways. “We can include an accelerometer, so the device knows when the user is going out to exercise. But usually when people start to exercise it's already too late.”
Improving performance at this stage is as much about behavioural science as algorithms and technology. “We are connecting the CamAPS FX app to smartwatches, so people can see the data more easily. Even something like adding dark mode to the smartphone app means people look at the information more.”
And there's the need to notify the regulator of major changes, which inhibits the ability to experiment. “CamAPS FX is a class 3 device, the same classification as an implantable defibrillator. Any small change brings a regulatory burden.”
There are legal obstacles too. “There is a lot of intellectual property in this area. We need to be super careful. There are four brands in the closed-loop sector, with the other three from the US. We need to make sure we are legally clear with our ideas.” This means the usual iterative approach with software is not possible. Fail fast and break things is rarely advisable in the medical field.
The policy hurdle
A theme which surfaces again and again in our interview is the role of policy-makers in restricting the roll-out of the artificial pancreas. Funding constraints mean those who could benefit, often don't. This in turn lowers the revenue stream of the tech firms involved, hindering development.
Professor Hovorka explains: “We can see two 25-year-olds with the same bad glucose control. One of them can get a closed-loop system, and the other cannot. Why? It's policy. We have a solution for type 2 diabetes, but the healthcare systems, apart from a few such as partially in France and partially in Australia, do not allow insulin pumps to be reimbursed in these populations.”
The scale of diabetes makes the issue profoundly political. In the UK alone 4.7 million people have diabetes, with 90% or 4.2 million with type 2. Diabetes UK estimates an additional 1.3 million people have undiagnosed type 2. “The cost is somewhere around £4,000 for a pump and accessories,” says Professor Hovorka. Running costs can be a few thousand pounds a year. This adds up to big numbers for the UK alone. But the return on investment is absurdly favourable.
“Look at the benefits,” says the professor. “Long-term there are health benefits. And the money saved by investing in expanding access to closed-loop systems would be substantial.” According to Diabetes UK, more than half of the £10.7bn spent annually by the NHS on diabetes related disorders is associated with complications, rather than routine care. Poor management of diabetes can lead to heart disease, nerve damage, macular oedema, and in extreme cases gangrene and amputation. The cost for a patient with secondary complications balloons far beyond the cost of a closed-loop system.
“It's a policy problem,” says Professor Hovorka, with exasperation. “There is concern about the costs if they open the gates to everyone.”
In the UK the term “treasury brain” describes the habitually short-term focus in HM Treasury, irrespective of government, a mindset denounced by figures such as the chief economist of the Bank of England. Professor Hovorka is frustrated by the funding formula he sees in the UK and internationally.
Cystic fibrosis, for example, is a neglected niche. CF patients can develop diabetes because the disease damages the pancreas, reducing insulin production. Professor Hovorka notes: “A colleague of mine at Cambridge is researching closed-loop for diabetes in cystic fibrosis patients. The current guidance does not allow them to use closed-loop, although CFRD, cystic fibrosis related diabetes, has a similar requirement to type 1. They could use it, but they don't. It's policy.”
Access varies by geography. NHS England is rolling out closed-loop systems nationally from 2024 over a five year period. Eligibility varies due to local services varying in staffing, training, and capacity. In Scotland funding from 2026/27 will be via individual health boards. Diabetes UK warns this could increase geographic variation in access.
“It's a huge postcode lottery,” laments Professor Hovorka. “We've had people on a closed-loop system during our trials. When it ends they ask to continue, only to be told it's not possible.”
Training is another obstacle. “It takes time for people to adapt to a new technology,” he observes. “It's about changing momentum and systems. In certain areas there's a wait time for 12 to 18 months for people with type 1, and that is just down to the speed of the infrastructure.”
“Sorry for talking about policy,” he adds. “Rather than the science.”
Fortunately for people living with diabetes, Professor Hovorka and his team are as motivated as ever to keep researching. “My colleague Dr Charlotte Boughton just finished a study in diabetes in cystic fibrosis, and since the study was conducted two of the patients died. We also see the fantastic cases. We had some newborns with type 1 diabetes that I put onto CamAPS straight away from day two or three. They use diluted insulin. And this makes the life of the newborns, and their mums and dads, better. Bearable.”
He says people with diabetes should take heart in the competition in the sector, with four companies, including his, producing control systems for the sensors and pumps. “Competition is a good thing for patients!” he says. “It's helpful to bring in more solutions. Penetration in the UK is only 20% of diabetics with type 1, so we've still got 80% not using a closed-loop system.” His aim for CamDiab is 30% of the world market, which is big enough for all market participants to prosper.
Professor Hovorka's message is clear: the artificial pancreas is here, and brings immense health benefits. He's done his work. Now he needs policy makers to play their part.
“Until a cure is found, and hopefully it will, my view is type 1 diabetes should be treated as much as possible with automated insulin delivery. Today we need to persuade healthcare professionals and the system to fund it.”
Susannah Dragosavac, Partner and Patent Attorney at Mewburn Ellis comments:
"It’s wonderful to see the progress Professor Hovorka is making in automated insulin delivery (AID) systems. AID systems can be life-changing for people living with type 1 diabetes, enabling better glucose control and allowing many to sleep through the night without having to actively manage their diabetes. The launch of a fully closed loop system in a commercial platform is an exciting advancement – I look forward to seeing how this fully closed loop system is adopted by users in their daily lives."
Written by Charles Orten-Jones.
.png)