According to the International Diabetes Federation’s latest Diabetes Atlas, an estimated 589 million adults worldwide now live with diabetes, and more than 250 million cases remain undiagnosed. Type 1 diabetes alone now affects close to 9.5 million people globally, up 13% since 2021.1International Diabetes Federation, IDF Diabetes Atlas, 11th Edition (2025) – diabetesatlas.org/resources/idf-diabetes-atlas-2025/2University of Washington Department of Medicine, “Over 250 million people worldwide unaware they have diabetes” (2025) – mednews.uw.edu/news/2025-IDF-diabetes-atlas3Diabetes Research and Clinical Practice, “Global type 1 diabetes prevalence, incidence, and mortality estimates 2025” – diabetesresearchclinicalpractice.com
Type 1 diabetes (5-10% of cases) is an autoimmune disease that leads to the destruction of insulin-producing pancreatic beta cells. It is characterized by insulin deficiency and resultant hyperglycemia. Type 2 diabetes (90-95% of cases) is characterized by insulin resistance and relative insulin deficiency.4American Diabetes Association; 2. Classification and Diagnosis of Diabetes: Standards of Medical Care in Diabetes—2021. Diabetes Care 1 January 2021; 44 (Supplement_1): S15–S33.
In this article, Alcimed explores the impact of artificial intelligence on diabetes care, through 4 different areas of application: clinical decision support, self-management tools, diabetes prediction and risk stratification, and diabetic retinopathy screening.
Lorenzo, Health Consultant at our office in Italy, breaks down in this short video the impact of artificial intelligence on diabetes care, through 4 areas of application.
AI algorithms can be used as decision support systems for diabetes management. The algorithm analyzes historical health data, including blood glucose levels, insulin dosages, dietary habits, and physical activity, to provide personalized insulin recommendations and treatment adjustments. It identifies patterns and offers evidence-based guidelines for maintaining optimal blood glucose levels.
An example is Advisor Pro, from Dreamed diabetes, that receives data from CGMs and/or insulin pumps, the data is analyzed using the MD Logic algorithm, searching for highs and lows along with specific individual patterns. Advisor pro produces recommendations including optimization of basal rate, carbohydrate ratio, correction factor (insulin sensitivity) and personalized management tips. The recommendations are reviewed and edited by the endocrinologists and are then shared with the patients.
This combination of AI-driven insights and expert validation helps patients make informed decisions, even in emergency situations, enhancing overall diabetes care.
Since this article was first published, Dreamed ‘s platform also received FDA clearance for type 2 diabetes, extending its AI-based, personalized insulin-dosing support beyond type 1. More broadly, generative AI and LLM are now being tested as clinical decision support tools: recent studies show they can generate personalized treatment suggestions from patients data. However, one study found such models less reliable at refining an existing plan, meaning that human oversight remains essential for treatment decisions.1“FDA greenlights DreaMed’s AI-backed insulin management tool for folks with Type 2 diabetes” – mobihealthnews.com2iScience, “Generative artificial intelligence in diabetes healthcare” (2025) – pmc.ncbi.nlm.nih.gov/articles/PMC12284359/3Clinical Diabetes, “Assessment of Large Language Models for Enhancing Diabetologist-Developed Personalized Treatment Plans” (2025) – diabetesjournals.org/clinical
AI algorithm can interpret biometric data from CGMs (Continuous Glucose Monitoring) and alert the patient and/or directly make adjustments to the blood glucose, by administrating insulin through an insulin pump.
The MiniMed 780G insulin pump system, combined with the Guardian Sensor 3 (CGM) by Medtronic, uses advanced AI algorithms. The data is collected by the CGM and is then analyzed by the MiniMed to predict glucose trends. The insulin pumps adjusts the insulin delivery based on these AI predictions, with the ultimate goal to keep the blood sugars in the optimal range, reducing manual adjustments by the patient and preventing episodes of hyper- and hypo- glycemia. Overall, the combined system offers a more precise and automated management of the disease, enhancing glucose control and reducing diabetes burden.
Medtronic’s MiniMed 780G system has kept expanding its AI-driven automated insulin delivery. In September 2025, the FDA cleared the system for adults with type 2 diabetes and approved integration with Abbott’s Instinct sensor, a smaller CGM offering up to 15 days of wear. A 2025 software update also refined automatic correction dosing, improving the accuracy and control of this closed-loop insulin-delivery technology.4Medtronic Newsroom, “FDA Clears MiniMed 780G System to Enable Integration with the Instinct Sensor… and Approves Use in Type 2 Diabetes” (Sept 2025) – news.medtronic.com
AI can be exploited to predict an individual’s risk for developing type 2 diabetes. The early stage identification of high risk patients is crucial to avoid life-threatening complications and to implement personalized prevention strategies.
Researchers at the United Arab Emirates University Intelligent Distributed Computing and Systems (INDUCE) laboratory are conducting large-scale tests of different AI models. The main goal is to identify the specific risk factors, such as genetics, lifestyles and different pre-existing life conditions, by applying advanced machine learning techniques on large datasets.
The research contributes to move towards early diagnosis and personalized prevention strategies, reducing the risks of type 2 diabetes.
Diabetes prediction models remain under active development. Researchers are exploring how generative AI can strengthen risk stratification and forecast complications by simulating patient data, addressing the data scarcity that has limited predictive accuracy. With roughly one in four adults with diabetes worldwide still undiagnosed, such tools could support prevention and more personalized care.5Annals of Translational Medicine, “Generative artificial intelligence in predictive analysis of diabetes and its complications” (2025) – pubmed.ncbi.nlm.nih.gov/41211117/
Learn more about how our team can support you in your projects related to AI in healthcare >
Diabetic retinopathy, the leading cause of blindness in the US, can be prevented with early detection. AI diagnostic algorithms are based on image processing and machine learning techniques to analyze retinal images and identify signs of diabetic retinopathy. The effectiveness is comparable to screening with in-office clinical exam by an eye care provider.
LumineticsCore is an example of Al diagnostic system that autonomously diagnoses patients for diabetic retinopathy. It analyzes the images quickly and provides results in less than a minute without the need of highly trained operators. Therefore, it can be integrated in the workflow of primary care seamlessly. LumineticsCore enhances accessibility to crucial eye care for diabetic patients, reducing the screening burden and ultimately helps to prevent vision lost.
Real-world evidence for AI retinopathy screening has grown. A 2025 study of LumineticsCore in a largely non-white patient population confirmed it remains a reliable screening tool. Further, LumineticsCore now also competes with systems like EyeArt, yet national data show fewer than 5% of patients with diabetes receive AI-based retinal screening, showing that adoption, not just diagnostic performance, remains a key challenge.6Southern Medical Journal, “Evaluation of the Efficacy of Artificial Intelligence-Based Screening for Diabetic Retinopathy in a Predominantly Non-White Population” (2026) – pmc.ncbi.nlm.nih.gov/articles/PMC13290037/7Modern Retina, “Study examines the use of AI-based detection of diabetic retinopathy in the US” (2026) – modernretina.com
As AI is becoming more and more established in the healthcare world, it has a great potential in diabetes care. The objective remains to reduce the burden of diabetes, by empowering patients, giving them access to advanced technologies and daily support. From generative models supporting clinical decisions to more accurate screening tools, these applications show how fast AI is reshaping diabetes medicine, though closing gaps in diagnosis and care access will matter as much as further technological development.
At Alcimed we keep exploring the fast-paced developments in AI in healthcare and can support you in your projects related to AI and diabetes care! Don’t hesitate to contact our team!
About the authors,
Gea, Senior Consultant in Alcimed’s Healthcare team in Italy
Annelisa, Consultant in Alcimed’s Healthcare team in Italy
Hélène, Senior Project Manager in Alcimed’s Healthcare team in Italy
Healthcare
T2D prevalence is rising at 2.5% per year and 60% of the diabetes patients are Asian. Discover 3 diabetic food innovations used to glycemic control in Asia.
Healthcare
The ever-increasing prevalence of diabetes, the associated costs to healthcare systems, and the growing regional disparities in care are all issues that we have to face. For patients with diabetes, ...
Healthcare
Continuous monitoring allows early detection Traditionally, clinical medicine relies upon “snapshot” measurements captured at singular or limited timepoints, assisting in determining a final ...