Continuous Glucose Monitoring with AI-enabled Predictive Algorithms for Diabetes Care: Expert Views from Hong Kong

This is a verbatim transcript of an interview conducted with Dr Elaine Chow in October 2025. The transcript has been lightly edited for clarity.

What significant shifts or insights have you observed in how diabetes is managed today?

So we live in very exciting times. In the past decade, we’ve seen many new drugs especially GLP-1 receptor agonist-based therapies and also SGLT2 inhibitors, which has really transformed the way we manage our diabetes and cardiovascular diseases. At the same time, we’ve seen advances in glucose monitoring technologies especially Continuous Glucose Monitoring or CGM.

The main challenges I see especially in Asia, is that of young onset type 2 diabetes. Currently in Asia, nearly 1 in 5 individuals with diabetes is diagnosed under the age of 40. And this is characterised by very aggressive beta cell decline, to the point that many of them do need insulin treatment on top of their oral glucose lowering drugs. At the same time, they develop very aggressive cardiorenal complications and they tend to progress faster than people with usual onset type 2 diabetes.

Currently in my clinic, I look after a lot of young onset type 2 patients and most of them are on insulin and also multiple daily injections of insulin with both basal and prandial insulin. So this is where I see that use of technology and CGM might be particularly useful. Currently most research studies on use of CGM is predominantly in people with type 1 diabetes, and I believe there’s a lot more work that needs to be done on optimising technology use in type 2 diabetes especially in Asia.

What are the main challenges encountered by patients in daily diabetes management?

So I think one of the main challenges is the sheer mental load of self managing the diabetes, especially for young adults with type 2 diabetes. Many of them are on multiple insulin injections and other therapies. And understandably, it’s actually very challenging as a young person live with diabetes, in their 30s. They might be juggling their personal lives with young families, having to look after old parents, at the same time, maybe breadwinners for their families and have to work. For people on multiple daily injections of insulin, they have to think about their food choices. So they may or may not have regular meal times or regular food intake. And every day they need to make decisions about their insulin and their food, 3 to 4 times a day, 365 times a year. And it is unsurprising that, many of them experience high level diabetes distress.

So what I think is really important is how we can help our patients reduce this decision fatigue and how we can minimise the mental burden, that they have to live with every single day. So sometimes my patients will ask me, why should I use a CGM? And I would tell them that with CGM, it gives you a tool that you can see what is happening to your glucose every day in response to your lifestyle or your drugs or the insulin that you’re taking. And with the CGM, in effect, you’re your own doctor every single day. So I really believe that using technologies and also with the help of AI predictions, I can truly help transform and simplify, and facilitate self-management of diabetes.

How would emerging diabetes technologies, particularly those incorporating advanced predictive algorithms, help address some of these unmet needs?

So I think use of AI and predictive algorithms can be very useful to help people with diabetes and make sense of the glucose data. Currently with CGM or the BGM, they might see a high or low glucose value and this might evoke a negative emotional response. But how can we actually help the person living with diabetes translate this information into timely actions? I really believe that’s the power of AI and predictive algorithms, so that we can be proactive about how we manage the glucose rather than being reactive.

So I’ll give you an example. I look after a lot of young type 2 patients in my clinic. And this one particular lady, she is a very busy manager, in her 30s. She has got type 2 diabetes, but she still has a suboptimal control with the HbA1c about 8%. And she really wants to get pregnant. So she currently is on multiple daily injections of insulin. And her problem was that, usually she runs quite a high glucose of up to 15 mmol/L late morning and after her breakfast. And yet, she is very worried about going low, having a hypoglycemia during her work meeting which may lead to embarrassment.

So with the help of CGM at the glucose prediction, she’s now able to see how her glucose will behave two hours later. So for example, she now realises it’s mainly her congee or her porridge, which is really incurring the very high postprandial surge. So she switched herself to a lower carb, bread option. And she’s now giving her appropriate dose of insulin about eight units. And now with her glucose prediction, she knows that she will be warned about impending hypoglycemia, so she doesn’t need to worry. Working with her in the clinic, we really saw a huge improvement in her HbA1c going from 8.2% down to 7.2%. And also her time-in-range really improved from about 50s up to the 82%. And I think most importantly, she felt a lot more empowered. And I would say she is now sailing and in control of her diabetes. I think CGM together with predictive algorithms can really help people take care of their anxiety around diabetes. And I think this can really help translate into better, longer term outcomes.

Which strategies are most crucial for optimal glycemic control and long‑term complication prevention in young‑onset type 2 diabetes?

So with young onset type 2 diabetes, early detection and intervention is most important. I think we need to think about multifactorial risk factor management. But in reality, the two most difficult things to control is number one, glucose, and secondly, weight. We’ve now have a lot of randomised controlled trials of CGM. And if you look across the data, the physician guided treatment titration actually plays a relatively small part. And what you can see both in trials and in real life is that the real impact that CGM makes is how you can support the patient, person with diabetes, better manage their day to day lives. And you can see the improvements in glucose already within the first 2 to 4 weeks.

So the question is, how can we make best use of these technologies, to help the young person with diabetes. So, I think it’s very important to help our people to onboard the technology. So for this, you need a multidisciplinary team. I would involve a clinician, diabetes educator, dietitians, nutritionists, but also their families and carers. And we need to get them on board to understand the value, the basic principles of diabetes self-management. And once they’re set up, I believe they’re more or less on a copilot. They would be navigating using the technologies every single day. And I think with the data that the CGM provides to you, which is much higher granularity, it actually opens up as a window to give us better insight in the day to day lives. And as a clinician, better understand what is happening to their diabetes every single day. So I believe with the power of CGM with AI and also these newer predictive algorithms, we can really help our patients and set them up with success in the very long run.

The views and opinions expressed by Dr Elaine Chow are her own views and opinions. Roche disclaims all liability in relation to these views and opinions.