Last week, Roche unveiled its CE-marked integration of its Accu-Chek SmartGuide continuous glucose monitoring (CGM) system with its mySugr app.

This innovation is designed to provide more predictive insights, reduce alarm fatigue, and ultimately improve diabetes care across Europe and beyond. The advancement underscores a growing trend in digital health of using data-driven technologies not only to collect information but also to translate it into actionable guidance for patients and healthcare providers.

The clinical implications of artificial intelligence (AI) enhanced CGM are significant. For decades, traditional blood glucose monitoring relied on fingerstick testing, which offered only snapshots of glucose levels. CGM technology changed that by allowing patients and clinicians to see trends, variability, and the impact of daily choices in real time. Now, with AI integrated into these systems, users can benefit from predictive alerts, such as warnings about impending hypoglycemia, before the event occurs. Such foresight can empower patients to take preventive steps and give clinicians richer data for tailoring treatment plans.

To understand how these developments are shaping clinical practice, Drug and Device World spoke with Dr. Jackie Elliott, Clinical Lead for Diabetes at Sheffield Teaching Hospitals. Dr. Elliott, who has worked extensively with CGM in both type 1 and type 2 diabetes populations, shared her perspective on how CGMs are transforming patient decision-making, reducing the burden of complications, and integrating into healthcare systems. This Q&A was conducted following Roche’s EASD 2025 presentation.

This interview has been edited for clarity, consistency, and length.

Phalguni Deswal [PD]: How have CGMs and now predictive insights changed decision-making compared to traditional fingerstick methods?

Dr. Jackie Elliott [JE]: It’s completely different. Fingerstick testing gives you isolated snapshots, but CGM provides a continuous picture that reflects real life. To get something similar with fingersticks, people would need to test six to eight times a day, and even then, the data would be incomplete. People often choose not to test when they suspect their glucose is high, because they don’t want that recorded on their download.

CGM overcomes all that. I had a young man with newly diagnosed diabetes who had a hypoglycaemic episode while walking to a football match. The CGM trace helped him understand that exercise, even just a mile and a half uphill walk, had a big impact on his glucose. That conversation gave him confidence that exercise was safe, as long as he adjusted insulin or meal timing. Without CGM data, I could never have had that teaching moment so early in his journey. Patients often describe CGM as “life changing,” and I agree. It allows both patients and clinicians to make decisions based on a complete and honest picture.

PD: For patients, interpreting CGM data can be overwhelming. How do you guide them?

JE: Some patients look at their data and instinctively know what to do. For others, it takes teaching. I often explain that if a hypoglycaemic episode happens at 1 am, it’s likely linked to evening meal insulin, while a 5 am hypo points to basal insulin. Once patients understand that patterns have explanations, they gain confidence.

The predictive insights are especially valuable. They warn patients before they actually go low, when they’re still in a calm state and can make rational choices—say, taking 15 grams of carbs. Contrast that with being in the middle of a hypoglycaemic episode, when adrenaline is surging, the heart is racing, and it’s almost impossible not to over-treat with too many sweets or juice. Avoiding that cycle is a major benefit, both for safety and for weight management.

PD: CGMs are often associated with insulin therapy, but do you see broader applications?

JE: Definitely. CGM has huge potential beyond type 1 diabetes. For people with type 2 diabetes, whether they’re on tablets, insulin, or even just lifestyle interventions, it can be incredibly powerful. I recently gave a CGM to a nurse with type 2 diabetes. She saw, first-hand, how activity after eating reduced her glucose spikes and how small lifestyle adjustments had a big impact. She’s since lost four stone, feels fitter, and even manages her workload better.

For many patients, CGM isn’t just about numbers, it’s about motivation. Seeing how your choices affect your glucose in real time can be the nudge people need to eat differently, move more, or question whether their current medication is right for them. It gives ownership back to the patient.

PD: Alarm fatigue is a real concern with CGMs. How should patients manage this?

JE: Alarm fatigue is real, and it can discourage patients. That’s why I always start with only low-glucose alerts. The priority is safety; catching hypoglycaemic episodes early prevents severe lows and reduces the amount of treatment needed. Interestingly, some patients even lose weight because they no longer need to consume excess calories to treat repeated hypoglycaemic episodes.

Once a patient is stable, I’ll consider adding high-glucose alerts, but carefully. There’s no benefit in setting an alert at 10 mmol/L, because it would trigger after almost every meal. I prefer to set a higher threshold, around 15–16, so it only goes off once in a while—usually when something has truly gone wrong, like forgetting a bolus. One alert a day is manageable; multiple alerts become noise.

From what I’ve seen, Roche has thought carefully about this issue. Their predictive alerts are designed to minimize unnecessary alarms while still offering meaningful, actionable warnings. That balance is crucial to long-term adoption.

PD: From a healthcare system standpoint, how can CGM reduce the long-term burden of diabetes?

JE: The potential here is enormous. CGM gives clinicians a population-level dashboard. In our clinic, we flag anyone who spends more than 10% of their time below range. That means each week we can identify and prioritize the 10–15 patients who need urgent attention. With fingersticks, that level of oversight would be impossible.

CGM also enables proactive care. Some patients don’t realize they’ve lost weight and therefore need less insulin, while others may have turned off alarms and require support in turning them back on. With CGM, we can spot those patterns quickly and intervene. That’s a far better use of limited healthcare resources than blanket check-ins with everyone.

Ultimately, by preventing recurrent hypoglycemia, improving adherence, and encouraging healthier behavior, CGM could reduce hospitalizations and complications. And in overstretched health systems, that efficiency is just as important as the individual patient benefits.