The 3rd Diabetes Digital Management Conference (DDMC 2026) Concludes Successfully, Launching a New Journey in Full-Lifecycle Diabetes Management

The prevalence of diabetes in China remains high, driving a continuous paradigm shift toward precision and full-lifecycle management. As digital technologies accelerate their integration into clinical practice, innovative tools such as Continuous Glucose Monitoring (CGM) and AI-driven clinical decision support systems are empowering clinicians to precisely identify glycemic fluctuations and risk factors. This evolution marks a transition from experience-driven to data-driven management, expanding the methodologies available for individualized, granular diagnosis and treatment.

On May 30, 2026, the 3rd Diabetes Digital Management Conference (DDMC 2026) officially convened in Shanghai. Hosted by the Institute of Active Health Strategy and Development at Shanghai Jiao Tong University, the conference adopted the central theme "Digital & Intelligent Empowerment · Global Collaboration: A New Journey in Full-Lifecycle Diabetes Management." The event brought together leading domestic and international endocrinologists, clinical scholars, and industry practitioners. Discussions focused on the frontiers and innovative applications of digital diabetes management, addressing critical areas such as building precision glycemic control systems, implementing clinical digital tools, and upgrading full-lifecycle care to jointly accelerate the digital transformation of diabetes management. This report highlights the key proceedings of the conference.

Empowering Primary Care: Overcoming 4 Core Challenges in Diabetes Management

Professor Weiping Jia from the Institute of Active Health Strategy and Development at Shanghai Jiao Tong University systematically analyzed the current state of diabetes prevention and control in China, outlining the key bottlenecks and strategic solutions for digital diabetes management in primary care settings. Professor Jia noted that China faces a high diabetes burden with severe chronic complications. Although primary care facilities represent the frontline of defense, they suffer from low awareness, treatment, and control rates, with the comprehensive control rate for blood glucose, blood pressure, and lipids standing at a critical 4.4%. Currently, digital management in primary care faces four core challenges:

l A lack of large-scale, longitudinal, multimodal, high-quality specialized disease databases based on the Chinese population.

l The "black box" nature of AI models, which lacks clinical interpretability.

l A deficiency in high-level clinical evidence-based data.

l Insufficient clinical applicability and low acceptance of AI deployment at the primary care level.

To address these pain points, Professor Jia’s team developed the "Deep Series" projects, establishing an end-to-end clinical deployment solution:

1. Ensuring Data Quality: Utilizing hierarchical clinician diagnosis and real-world data training to resolve model degradation and hallucination risks caused by AI-generated data.

2. Enhancing AI Interpretability: Breaking through the interpretability bottleneck through the quantitative analysis of retinal vascular parameters and heatmap visualization—achievements highly recognized by the international diabetes community.

3. Validating Clinical Benefits: Conducting prospective real-world studies to confirm clinical efficacy. The Deep Series AI systems enable precise risk stratification and management for diabetic retinopathy, diabetic kidney disease, and stroke, significantly reducing the risk of complication progression while improving patient self-management compliance.

4. Facilitating Technical Deployment: Achieving real-world adoption and continuous model optimization through user-friendly interface design and ongoing training for primary care physicians. This management framework has been integrated into China's National Guidelines for the Prevention and Control of Diabetes in Primary Care.

Professor Jia emphasized that digital diabetes management in primary care must adhere to the principle of "human-AI collaboration" to prevent the devaluation of clinical expertise. AI should serve strictly as a clinical assistant tool; all diagnostic and treatment plans must be verified and executed by physicians to ensure medical safety, thereby establishing a reproducible and scalable digital management model for primary care.

International Perspectives: Real-World Evidence and Digital Health Driving Innovation

Professor Julia Mader from the Department of Endocrinology and Diabetology at the Medical University of Graz shared European clinical practices, illustrating individualized management pathways for type 2 diabetes (T2DM) powered by real-world data and artificial intelligence. She noted that CGM is a highly accepted management technology among patients in Germany, Austria, and Switzerland, with penetration rates steadily rising to cover full-course management from prediabetes to late-stage complications.

Type 2 diabetes is a progressive disease requiring personalized adjustments of glycemic targets based on disease stage, age, and overall health status. With the widespread adoption of AI, its integration with CGM can support individualized clinical decisions, data interpretation, glycemic fluctuation prediction, dietary carbohydrate estimation, and adjunctive medication management. Professor Mader emphasized that the "CGM+AI" combination will be a vital driver for future diabetes care. Rather than replacing clinicians, it empowers them to deliver more precise, proactive, and patient-centered individualized treatment.

Professor Lee-Ling LIM from the Department of Medicine, Faculty of Medicine, University of Malaya, connected global chronic disease prevention trends with the goals of the Healthy China 2030 Blueprint, sharing the deployment pathways and core value of digital health technologies. She pointed out that digital health now spans full-lifecycle diabetes scenarios—including blood glucose monitoring, insulin delivery, diet and exercise management, telemedicine/telemonitoring, and health alerts—serving as a core enabler for chronic disease control. However, effective diabetes care still faces multifaceted challenges, requiring an upgrade from "standalone apps" to an "all-encompassing digital ecosystem."

Leveraging the "ABCDEF Data-to-Action" framework, with a "Diabetes Registry System" and "Electronic Medical Record (EMR) Integration" at its core, this approach links clinicians and patients to drive multi-metabolic target compliance. The goal is to build an ideal patient-centered ecosystem supported by pharmacies, health insurance, medical devices, consumables, and regulatory authorities.

In terms of technical innovation, Professor Lim elaborated on the evolutionary path of CGM from innovative exploration to clinical implementation, the functional transition of wearables into health monitors, and the clinical prospects of AI predictive analysis. She also introduced DeepDKD, a deep learning system based on retinal imaging designed for the screening and differential diagnosis of diabetic kidney disease. Industry data indicates that the global self-monitoring technology market grew fourfold between 2020 and 2026, with the Asia-Pacific region accounting for 23.3%. CGM, non-invasive monitoring, and AI predictive analytics represent the core directions for future innovation. 

Deploying Digital & Intelligent Diabetes Management: Agentic Innovation, Hospital-wide Glycemic Models, and Multi-Scenario CGM Validation

Dr. Jiyun Zheng, Vice General Manager and Board Secretary of Sinocare, shared insights on building specialized disease agents anchored on trusted data spaces to address clinical AI deployment bottlenecks. She noted that while core AI capabilities have advanced—enabling the scaled application of clinical-grade AI knowledge platforms and generative AI summaries to reshape public health information access—the challenge of "AI hallucination" (including data distortion, fabricated information, and sycophantic bias) remains a critical bottleneck restricting deployment in medical scenarios. Coupled with gaps in public health literacy, these hallucinations amplify the clinical risks of misinformation.

To address this challenge, iCan Health (Intelligent Edition) launched "Xiao Nuo Agent," a conversational health management companion. Built upon the iCanNet trusted data space, Xiao Nuo Agent is grounded in the predictability, computability, and stability of real-world data science. It integrates multi-dimensional patient data—including home monitoring, lifestyle behaviors, and personal health records—to filter out distorted information at the source, resolving AI hallucinations from their root cause.

When facing complex blood glucose curves, patients can ask questions at any time to receive immediate professional interpretation and guidance, progressively enhancing their health self-awareness. Throughout continuous interactions, the system continuously learns the patient's behavioral habits and personal preferences. Relying on iCanNet to dynamically update health insights, it constructs a personalized health data profile for each user, outputting precise, individualized glycemic control plans through specialized clinical algorithms. 

Two Decades of CGM Research: From Technological Iteration to Comprehensive Clinical Frameworks

Professor Jian Zhou from Shanghai Sixth People's Hospital, reviewed the 20-year developmental history of CGM, systematically tracing its comprehensive breakthroughs across market evolution, clinical metric frameworks, and multi-scenario applications. Professor Zhou noted that The global CGM market grew from 200million in 2005 to14 billion in 2024, and is projected to surpass $30 billion by 2033, with the technology progressively expanding from professional clinical applications to the general wellness population.

Regarding core metric development, domestic research began in 2004 by uncovering glycemic drift patterns in individuals with normal glucose tolerance, establishing standard CGM reference values for the Chinese population in 2009. In 2018, Chinese researchers demonstrated for the first time that Time-in-Range (TIR) is independently associated with diabetic retinopathy, separate from HbA1c. This finding became the world’s first evidence-based validation supporting the clinical application of TIR and was subsequently adopted by international guidelines and textbooks. Furthermore, the Complexity Growth Index (CGI) of blood glucose time series can measure the body's homeostatic regulatory capacity and correlates with adverse diabetes outcomes, emerging as a promising new early indicator for evaluating glycemic homeostasis.

In terms of clinical application, multiple randomized controlled trials (RCTs) have confirmed that CGM significantly improves glycemic control in patients with T2DM. Within hospital settings, China has established CGM-based inpatient glycemic management systems and developed mobile ward-round platforms. The multicenter prospective INDIGO-ICU study has further extended the application of CGM into intensive care units. Additionally, research has identified specific CGM metrics associated with the occurrence and adverse outcomes of gestational diabetes mellitus (GDM), driving international multicenter, outcome-based analyses to define diagnostic thresholds and explore the potential of CGM in GDM diagnosis.

In the domain of AI-CGM fusion, the GluFormer model delivers precise predictions of future glycemic trends based on CGM datasets. Individualized precision phenotyping based on carbohydrate responses has revealed pronounced interpersonal variations in postprandial glycemic responses, highlighting the necessity of personalized dietary interventions. Furthermore, combining wearable devices with routine blood panels yielded an Area Under the Curve (AUC) of $0.80$ in predicting insulin resistance.

Professor Zhou summarized that CGM is transforming from a traditional "blood glucose monitoring method" into a "diabetes phenotypic characterization tool," expanding its reach into non-diabetic populations and serving as the central lever driving the advancement of diabetes management technologies.

Digital Healthcare Debate: Does Clinical AI Enhance or Diminish a Physician’s Core Decision-Making Ability?

The conference featured a structured thematic debate centering on the motion: "Does the development of AI in intelligent diabetes diagnosis and treatment enhance or diminish a physician's core clinical decision-making ability?" The session was co-moderated by Professor Xiaomu Li from the First Affiliated Hospital, Zhejiang University School of Medicine, and Professor Haibing Chen from Shanghai Tenth People's Hospital.

The Affirmative team comprised Professor Haixia Liu (Weifang People's Hospital), Professor Yuan Hu (Wuxi Traditional Chinese Medicine Hospital), and Professor Hongyun Lu (Zhuhai People's Hospital). The Negative team consisted of Professor Xianjie Hu (Jining First People's Hospital), Professor Zhiqing Hu (Integrated Chinese and Western Medicine Hospital, Southern Medical University), and Professor Yi Zhang (The First Affiliated Hospital of Quanzhou, Fujian Medical University). The teams engaged in a rigorous analysis of the values and risks of clinical AI.

The Affirmative argued that AI significantly enhances core clinical decision-making: AI rapidly screens vast volumes of medical literature and imaging data, flagging suspicious lesions within seconds, liberating physicians from data sorting to focus on core clinical judgment. Furthermore, it compensates for experience gaps among primary care physicians; for example, in diabetic retinopathy screening, AI systems perform fundus analysis comparable to that of senior clinicians within minutes, effectively empowering primary care chronic disease management. AI also functions as an educational tool, providing real-time feedback to younger physicians and accelerating clinical competency. Additionally, AI connects fragmented laboratory examinations, follow-up visits, and health education, facilitating early risk detection and expedited decisions. The Affirmative emphasized that AI is an adjunctive tool rather than a replacement, noting that just as glucometers and CT scans did not diminish physician capabilities, AI drives the upgrading of diagnosis and treatment.

The Negative countered that over-reliance on AI diminishes core clinical decision-making: AI systems that directly provide diagnostic conclusions permit clinicians to bypass the complete process of clinical reasoning, leading to a potential loss of independent judgment among younger physicians. Currently, high-level clinical evidence proving that AI consistently outperforms physicians in complex, real-world scenarios remains insufficient. Crucially, the core values of medicine—trust, continuous longitudinal care, and professional accountability—are elements that AI cannot provide. The Negative added that the growth of young clinicians requires an inductive process of learning from errors, and foundational skills such as literature review and physical examinations cannot be skipped. They cautioned against the potential negative impacts of AI on residency training frameworks.

Following robust rounds of constructive arguments, rebuttals, and free debate, the Affirmative team was declared the winner. The consensus concluded that AI serves as an exceptional clinical assistant tool, provided the core principle of "physician-led, AI-assisted" is strictly maintained. Medical students and trainees should establish a complete foundation of clinical reasoning before integrating AI tools into their workflows to avoid over-dependence, ensuring they embrace technology while upholding basic clinical skills and humanistic medical care to truly enhance healthcare quality.

References

1. Meng Z, Lim LL, Jia W, et al. Lancet Digit Health 2025.

2. American Diabetes Association. Standards of Care in Diabetes—2026. Diabetes Care. 2026;49(Suppl 1):S1–S350.

3. Jendrike N, Link M, Öter S, et al. Performance of a New Continuous Glucose Monitoring System in German Adults Living with Diabetes. Diabetes Ther. 2026;17(2):287-299.

4. Yu M, Zhou J, Xiang KS, et al. Fluctuations of continuous glucose monitoring in subjects with normal glucose tolerance. National Medical Journal of China, 2004, 84(21):1788-1790.

5. Zhou J, Li H, Ran X, et al. Reference values for continuous glucose monitoring in Chinese subjects. Diabetes Care. 2009 Jul;32(7):1188-93.

6. Lu J, Ma X, Zhou J, et al. Association of Time in Range, as Assessed by Continuous Glucose Monitoring, With Diabetic Retinopathy in Type 2 Diabetes. Diabetes Care. 2018 Nov;41(11):2370-2376.

7. Cao L, Wang YX, Lu JY, et al. Research progress on the application of continuous glucose monitoring in non-diabetic populations. Chinese Journal of Diabetes, 2025, 17(05):623-627.