From mobile screening tools in Egypt and SMS coaching in India to offline chatbot systems in rural Italy, AI is redefining what is possible in diabetes care. These solutions are not futuristic fantasies—they are working now, often in the most under-resourced settings, and the evidence behind them is growing.
Access to basic healthcare remains a daily challenge in many parts of the world. For the millions of people living with diabetes in low- and middle-income countries (LMICs), the lack of diagnostic tools, trained professionals and consistent treatment options means that complications often arrive before diagnosis. However, artificial intelligence (AI) is beginning to offer a new kind of support—flexible, low-cost and increasingly effective.
The latest figures from the International Diabetes Federation (IDF) are sobering. An estimated 589 million adults worldwide are currently living with diabetes—around one in nine adults. Today’s projections say that number will rise to 853 million by 2050. Of these, 81% live in LMICs, where healthcare systems face ongoing constraints. With more than 252 million people—over 40% of all cases—living with undiagnosed diabetes, this flashing red light demands bold, accessible and scalable solutions.
Prohibitive costs, limited infrastructure and a shortage of healthcare professionals all contribute to the barriers people face in accessing consistent diabetes care. Many communities rely on fragmented services or must travel long distances for routine care. In these contexts, AI is not just a luxury of advanced healthcare systems—it is a pathway to equity.
Many communities rely on fragmented services or must travel long distances for routine care. In these contexts, AI is not just a luxury of advanced healthcare systems—it is a pathway to equity
So, what is happening at the crossroads of diabetes care and artificial intelligence? Around the world, we are seeing the emergence of innovative tools—not only in high-tech labs but in clinics, communities and households where resources are scarce and the need is greatest.
At the recent World Diabetes Congress 2025 in Bangkok, the session “Implementing Artificial Intelligence (AI) Solutions for Diabetes Diagnosis and Management in Resource-Limited Settings” looks at how cutting-edge technology is stepping in where healthcare systems cannot—bringing smart, scalable solutions to the frontlines of diabetes care. Dr Ambady Ramachandran, a long-standing advocate for diabetes prevention, led a study in India that used AI to tailor SMS messages promoting healthy behaviour among men with impaired glucose tolerance. The result: a 36% relative risk reduction in the progression to type 2 diabetes over two years.
India has one of the highest diabetes burdens in the world, with more than 212 million adults affected. In this context, small, simple interventions can make a significant impact. “We didn’t call it AI back then,” Dr Ramachandran said at the Congress. “But the system learned and adapted. It understood behaviour—and it changed it.”
The intervention has since been scaled nationally, with over 130 million SMS messages sent to hundreds of thousands of people. Among those identified with diabetes, 67% attended a follow-up within six months. These results underline how digital tools—especially those accessible through basic mobile phones—can transform outcomes quickly. Similar trials in South Africa and Malawi (StAR2D) and among younger adults in Australia (TEXT2U) reinforce the idea that simple digital messages, when well-designed, can improve care engagement.
Cutting-edge technology is stepping in where healthcare systems cannot—bringing smart, scalable solutions to the frontlines of diabetes care
Egypt, where over 1 in 5 adults (20-79 years) are living with diabetes, faces a similar challenge. Healthcare services in rural areas are often fragmented or out of reach. For Professor Inass Shaltout, the solution lies in making technology work for households.
She and her team developed a mobile screening app using AI-powered algorithms built around simple self-reported data—age, weight and family history. The app also offers education on nutrition, physical activity and risk management, all in Arabic and adapted to local cultural contexts.
“Obesity is a tsunami sweeping through younger generations,” she said. “Twenty percent of obese adolescents already have pre-diabetes. We need to meet them where they are—on their phones, with tools they trust.”
Initial studies show high acceptability, especially among families with adolescents. For many, it is the first time they have had access to any form of structured diabetes risk assessment.
When we think of low-resourced countries, Italy does not necessarily come to mind. But, in its mountainous and rural regions, people can still face significant barriers to diabetes care. Broadband is unreliable, travel is difficult, and the nearest doctor may be far away.
Dr Felice Strollo’s project, DIABETICA, is a conversational AI system that delivers personalised guidance on blood glucose control, diet and insulin use through SMS or local devices. It works offline, using natural language processing (NLP) and responding to people’s questions with empathy and clarity.
“AI won’t replace clinicians,” Dr Strollo explained. “But it can reach people the system doesn’t.”
In early studies, people found DIABETICA easier to understand than printed materials. It offered a sense of connection and real-time support. And this is all done without a live internet connection, which can make all the difference for older adults or those with lower health literacy.
When we think of low-resourced countries, Italy does not necessarily come to mind. But, in its mountainous and rural regions, people can still face significant barriers to diabetes care
Other AI-driven approaches are gaining traction in resource-limited settings:
These solutions are not theoretical. In multiple peer-reviewed trials and real-world settings, they are helping detect diabetes earlier, reduce complications and empower communities to take control of their health.
A growing body of research supports the effectiveness of AI in improving diabetes care. In Rwanda, the RAIDERS trial demonstrated that AI-assisted diabetes-related retinopathy screening significantly increased referrals to specialist services, suggesting real-world impact in underserved settings. In India, a multicentre study of the AIDRSS system reported 100% sensitivity for detecting referable retinopathy and boosted screening rates in rural areas.
Meanwhile, voice-based AI tools across sub-Saharan Africa and South Asia could be part of the solution. Trials in these regions have revealed their potential for community-led, non-invasive screening. In the United States, the ACCESS study found that autonomous AI-based eye exams were more effective than traditional referral methods for identifying retinopathy in young people. Although encouraging, these findings underscore the importance of real-world testing, clearly defined policies and responsible integration into local health systems.
Meanwhile, voice-based AI tools across sub-Saharan Africa and South Asia could be part of the solution. Trials in these regions have revealed their potential for community-led, non-invasive screening
Despite the promise of AI in transforming diabetes care, several challenges remain. One pervasive challenge is data bias. Because AI models often rely on information from high-income countries, they limit their relevance in more economically and culturally diverse settings. So, local validation has to be present to ensure accuracy and relevance. Infrastructure gaps also present barriers. AI tools often rely on smartphones, internet access or stable electricity, which may not be consistently available in remote or underserved areas. In these environments, offline or hybrid models are preferable and necessary.
Privacy is another concern. Laws that protect health data, such as the Health Insurance Portability and Accountability Act (HIPAA) in the US or the General Data Protection Regulation (GDPR) in the EU, which includes health data protection under broader privacy laws, differ widely from country to country. Clear systems for securing personal medical information are still underway in many regions.
Sustainability is one of the last hurdles. While pilot projects frequently show promising results, long-term success depends on sufficient funding, supportive policies and the ability to integrate and maintain these technologies within existing public health systems.
A recent scoping review also noted that most AI research in diabetes focuses on technical accuracy rather than real-world outcomes. For long-term success, implementation, cost-effectiveness and equity must be front and centre.
AI will not solve health system inequality on its own. Still, as these projects show, it can help fill critical gaps—especially when designed with local communities in mind.
Whether it is SMS coaching in Chennai or a chatbot in a mountain village in Calabria, the message is the same: AI can make care more accessible, more personalised and more human-centred. But for that to happen, these tools must be inclusive, safe and aligned with the needs of the people using them.
As the global diabetes community looks ahead, partnerships between researchers, clinicians, developers, and people with diabetes will be key. The future is not just about smarter algorithms—it is about smarter systems that serve everyone.
In a move that reflects the growing momentum in this space, the International Diabetes Federation (IDF) has established a new Technology and AI Working Group. Its aim is to better understand how emerging technologies can support diabetes prevention, diagnosis and care, particularly in diverse healthcare settings around the world. The Working Group brings together experts in digital health, clinical practice, data ethics and AI development to explore practical, people-centred solutions. By creating a platform for knowledge exchange and collaboration, the International Diabetes Federation hopes to shape more inclusive and practical approaches to digital health that serve everyone—no matter where they live. Learn more
