AI in diabetes care is already part of daily life for many people living with the condition. It reads the stream of numbers from a glucose sensor, adjusts insulin through a pump, and checks photos of the back of the eye for early signs of damage. None of it replaces the care team, but it changes how much of the work a person does by hand.
The scale of the condition explains the attention. The World Health Organization (WHO) reports that the number of people living with diabetes rose from 200 million in 1990 to 830 million in 2022, and that 14% of adults aged 18 and over were living with diabetes in 2022, up from 7% in 1990. More than half of adults aged 30 and over with diabetes were not taking medication in 2022.
This guide explains what the technology does at each step, where it falls short, and what to ask the people who look after you.
What diabetes care involves day to day
Diabetes is a long-term condition in which blood glucose stays too high, either because the pancreas does not make enough insulin or because the body cannot use insulin well. Insulin is the hormone that controls blood glucose. Over time, high blood glucose can damage blood vessels in the heart, eyes, kidneys and nerves, with blindness, kidney failure, heart attacks, stroke and lower limb amputation among the complications WHO names.
There are two main types. In type 1 diabetes the pancreas makes little or no insulin, and a person needs insulin every day; WHO recorded 9 million people with type 1 diabetes in 2017. In type 2 diabetes the body stops using insulin properly. More than 95% of people with diabetes have type 2, which WHO notes is often preventable through a healthy weight, regular activity, a healthy diet and not smoking.
Day-to-day care is a routine of small decisions: checking glucose, eating, moving, taking medicines, watching for signs of trouble. The American Diabetes Association (ADA) defines diabetes technology as the hardware and software people use to help with self-management, and notes no device works well without education, training and ongoing support.
Continuous glucose monitors turn readings into patterns
The most widely used tool is the continuous glucose monitor, or CGM. The US Centers for Disease Control and Prevention (CDC) describes it as a small wearable device that attaches to the stomach or arm and uses a tiny plastic tube to read glucose from the fluid between cells just under the skin. It sends a new reading every few minutes to a phone or receiver, and many models alert when glucose is rising or falling fast.
The difference from a fingerstick test is the pattern. A fingerstick gives one number at one moment. A CGM shows whether glucose is rising, falling or steady, and how it responded to a meal, a walk or a night's sleep. The CDC explains that results are often read as "time in range" — the share of the day glucose spends inside a target band, usually 70 to 180 mg/dL, with a goal of 70% of the day, about 17 hours.
Sensors need replacing every 7 to 14 days, and the CDC notes a fingerstick check is still needed at times, because readings from fluid under the skin can differ from blood readings. Results can also be shared with the care team.
The ADA recommends CGM be offered to people with type 1 diabetes early, even at diagnosis, and that device reports be standardised around a one-page summary and a weekly summary. In one trial the ADA cites, starting an automated insulin system within 21 days of a type 1 diagnosis gave 10% more time in range and a lower A1C after 12 months.
Hybrid closed-loop systems adjust insulin on their own
A hybrid closed-loop system joins the two parts together: a CGM, an insulin pump, and an algorithm between them. The algorithm reads the glucose numbers and tells the pump how much insulin to give, adjusting the dose through the day and night. It is sometimes called an artificial pancreas, though the person still has to tell the system about meals and exercise.
In England and Wales, the National Institute for Health and Care Excellence (NICE) recommended these systems in December 2023, in appraisal TA943. Diabetes UK reports the recommendation covers adults with type 1 diabetes whose HbA1c is 58 mmol/mol (7.5%) or higher, or who have disabling low blood sugar, despite the best care with a pump or a glucose monitor. It also covers all children and young people under 18, and people pregnant or planning a pregnancy.
The same guidance says the systems should only be used with the support of a trained team experienced in insulin pumps and glucose monitoring, and that people using them should be offered an approved education programme. NHS England agreed to offer the technology to eligible people in a phased programme over five years, starting in early 2024.
The evidence is that the systems help people with type 1 diabetes improve blood sugar levels, have fewer low blood sugar episodes, and find self-management easier, taking on small repeated adjustments through the day and night.
AI eye screening catches damage earlier
Diabetes can damage the blood vessels in the eyes, and WHO notes that it can cause permanent vision loss. Screening means checking the eyes regularly so damage is found before sight is affected — a check that has usually needed an eye specialist to look at photos of the retina.
In April 2018, the US Food and Drug Administration (FDA) authorised the first AI device that makes the screening decision on its own. The system, IDx-DR, reads photos of the retina taken with a special camera and returns one of two answers: more than mild diabetic retinopathy detected, so the person is referred to an eye care professional, or negative, so the person is screened again in 12 months.
The decision followed a trial of more than 900 adults with diabetes at 10 primary care sites. The American Academy of Ophthalmology reports that the system showed a sensitivity of 87.3% — the share of people with the condition it picked up — and a specificity of 89.5% — the share of people without it that it correctly cleared.
Because no eye specialist needs to read each image, a GP surgery or diabetes clinic can run the screening and refer only the people who need a specialist — which matters most where eye clinics are scarce.
Where AI in diabetes care falls short
The tools are real, but they do not reach everyone. WHO's figures show the gap: more than half of adults aged 30 and over with diabetes were not taking medication for it in 2022, and treatment coverage was lowest in low- and middle-income countries. The CDC notes that without insurance, a CGM can cost thousands of dollars a year.
Access is uneven even where the technology is recommended. A report by the Health Services Safety Investigations Body (HSSIB), covered by Nursing Times in November 2026, found that at the time of its investigation only 12% of eligible adults and 76% of eligible children and young people under 18 in England were using a hybrid closed-loop system. It described a "postcode lottery" in access: older people and people with learning disabilities or mental health problems were less likely to be offered the technology, sometimes because they were not aware of it, sometimes because of assumptions about whether they could use it.
Training is the other weak point. The same report found community nurses increasingly supporting patients who use diabetes technology without having received the training to help them, and the ADA stresses that prescribing a device must come with initial and ongoing education, including how to use the data and how to handle alarms.
There are limits in the technology itself. A CGM reads fluid under the skin, not blood, so its numbers can lag when glucose is changing fast. An algorithm can only act on the data it is given: it cannot see a missed meal, a faulty sensor or an infection coming. And the data these devices collect is health data, so how it is stored and shared matters as much as how accurate it is.
Frequently asked questions
Can AI in diabetes care replace my doctor?
No. The tools support care: they sense, flag and adjust within limits set by a care team. NICE says hybrid closed-loop systems should only be used with the support of a trained team, and the ADA says no device works well without education and ongoing support. Decisions about medicines still belong with a qualified healthcare professional.
Do I still need fingerstick tests if I use a CGM?
Sometimes, yes. The CDC says glucose readings from the fluid under the skin can differ from blood readings, so it is worth checking with a fingerstick at times, especially when starting out. Ask your care team how often to check.
Is AI eye screening as good as seeing a specialist?
It is a screening step, not a replacement for eye care. In its trial, IDx-DR picked up 87.3% of people with more than mild diabetic retinopathy and correctly cleared 89.5% of people without it. Anyone flagged is referred to an eye care professional.
AI in diabetes care is best understood as a set of helpers, not a cure. A sensor turns scattered readings into a pattern a person can act on. A closed-loop system takes on small insulin adjustments through the day and night. An eye-screening tool brings a check that once needed a specialist into an ordinary clinic.
If you live with diabetes, the useful questions are practical: what does your care team already offer, what would a sensor or a pump change in your daily routine, and what training and support come with it? If you work in health, care or community support and need clear, careful content readers can trust, see our services, or get in touch through our contact page.
What to watch next
Watch whether the NHS five-year rollout of hybrid closed-loop systems closes the gap the HSSIB report found — 12% of eligible adults using the technology at the time of its investigation. Watch whether CGM prices fall and coverage widens, since cost is still the main barrier. And watch for more AI screening tools following IDx-DR into ordinary clinics.
This article is for general information and is not medical advice. Please speak to a qualified healthcare professional about your own situation.
Sources
- World Health Organization — Diabetes — date checked: 2026-11-07 — https://www.who.int/news-room/fact-sheets/detail/diabetes
- Centers for Disease Control and Prevention — Continuous Glucose Monitors — date checked: 2026-11-07 — https://www.cdc.gov/diabetes/treatment/continuous-glucose-monitors.html
- American Diabetes Association — 7. Diabetes Technology: Standards of Care in Diabetes—2025 — date checked: 2026-11-07 — https://diabetesjournals.org/care/article/48/Supplement_1/S146/157557/7-Diabetes-Technology-Standards-of-Care-in
- Diabetes UK — NICE announces final appraisal recommendations on hybrid closed-loop systems for type 1 diabetes — date checked: 2026-11-07 — https://www.diabetes.org.uk/about-us/news-and-views/hybrid-closed-loop-technology-recommended-thousands-type-1
- American Academy of Ophthalmology — AI device for detecting diabetic retinopathy earns swift FDA approval — date checked: 2026-11-07 — https://www.aao.org/education/headline/first-ai-screen-diabetic-retinopathy-approved-by-f
- Nursing Times — Better training in new diabetes tech needed for community nurses — date checked: 2026-11-07 — https://www.nursingtimes.net/diabetes/better-training-in-new-diabetes-tech-needed-for-community-nurses-04-11-2026/
Eri-Tech Global Ltd
Eri-Tech Global Ltd is a digital solutions company working with small businesses and growing brands. Learn more at /about.
