Digital Management of Early-Onset Type 2 Diabetes: Empowerment, Challenges, and Future Outlook

DHT: Types and Functions

DHT encompasses mobile health (mHealth) applications, telemedicine and remote monitoring platforms, continuous glucose monitoring (CGM), and regulatory-approved digital therapeutics (DTx). These tools typically integrate functions including medication reminders, dietary and physical activity tracking, educational content delivery, online messaging/video follow-ups, and automated decision support [13].

DHT is transforming the management of EOT2D by improving empowerment and optimization of care, as shown in Table 1. Personalized education delivered through mobile/web apps, messaging, and online platforms enhances health literacy, while visualization of integrated data on cardiometabolic risk factors, diet, physical activity, and sleep patterns supports self-monitoring and adaptive behavioral change. DHT also strengthens empowerment via remote consultations, asynchronous communication, and peer support, facilitating shared decision-making and sustained engagement (Fig. 2) [14]. Compared with conventional care models, integration of DHT with electronic health records allows prioritization of high-risk people and more efficient monitoring, with evidence suggesting improved glycemic outcomes and favorable cost-effectiveness [15]. Notably, emerging data from EOT2D populations indicate that internet-based care, SMS-based support programs, and CGM-supported management can improve glycemic control, treatment adherence, and patient engagement, further underscoring the potential of DHT-enabled, integrated care models for people with EOT2D [16–18]. Taken together, these strategies highlight how DHT can address both the clinical intensity and life-course challenges of people with EOT2D. However, DHT is not a monolithic intervention. Its impact depends on the modality, level of integration, and intended function within the care pathways. The following section examines five distinct categories of DHT, outlining their strategies, evidence to date, and relevance to the management of EOT2D.

Continuous Glucose Monitoring (CGM)

CGM technology employs subcutaneous sensors to monitor interstitial fluid glucose concentrations, providing real-time dynamic glucose profiles [19, 20]. Compared with conventional fingerstick glucose monitoring, a 2024 meta-analysis of 12 randomized clinical trials reported a 0.3%–0.4% reduction in HbA1c and a 6% improvement in time-in-range (TIR) with use of CGM in people with T2D treated with insulin therapy and/or oral glucose-lowering drugs [19]. People with T2D who experience greater lifestyle variability and a higher risk of nocturnal hypoglycemia and hypoglycemia unawareness will likely benefit from CGM [21].

CGM allows personalized, real-time feedback that helps young adults visualize how diet, physical activity, sleep, stress, and medication adherence influence their daily glycemic support shared decision-making, and enable remote titration of therapies when CGM data are integrated with telemedicine platforms and electronic health records, thereby positioning CGM as a central “data spine” that underpins other DHT for people with EOT2D [22]. Emerging multimodal deep learning models that fuse real‑time CGM and clinical data can now forecast interstitial glucose 15–60 min ahead with clinically acceptable accuracy, creating a window for algorithms to anticipate glycemic excursions rather than merely record them [23]. For people with EOT2D, such predictive engines could be embedded into “AI dietitian”–style tools that deliver culturally-tailored prompts about diets, physical activity, and medication, shifting digital health from passive monitoring to proactive risk management across home, school, and workplace settings. To translate these “AI dietitian” systems into routine care, models will need training on diverse, high‑quality datasets, rigorous clinical validation, and localization to local foods, languages, and care pathways in close collaboration with clinicians and dietitians [24].

Notably, existing studies in EOT2D are limited by small sample sizes, with pilot randomized and feasibility studies reporting short-term improvements in glycemic profiles and acceptability of CGM among publicly insured or clinic-based EOT2D [18, 25]. This paucity of large-scale data is driven even more profoundly by the unique socio-demographic and behavioral barriers inherent to EOT2D. People with EOT2D often present with higher rates of socio-economic deprivation, mental health conditions, and competing life demands compared with those with usual-onset T2D (Fig. 1) [26, 27]. These factors culminate in high default rates. Taken together, the small sample sizes in DHT studies reflect not merely a limited epidemiological pool, but a highly vulnerable, “hard-to-reach” population that traditional diabetes care persistently struggles to engage.

While initial uptake of CGM may be encouraging, sustained use often declines over time because of device burden, skin issues, costs, and access barriers [25]. These observations underscore that, although CGM can provide the data backbone for many DHT applications, specific strategies are needed to support long‑term acceptability and continuous use in people with EOT2D [25]. In a population facing long disease duration and elevated lifetime risk of cardio-renal events, such continuous, context-rich monitoring offers a plausible pathway to better cardiometabolic risk factor control and clinical outcomes, although head-to-head comparisons of CGM efficacy and hard endpoints (hospitalization, cardiovascular and kidney events, mortality) between people with EOT2D and usual-onset T2D remain limited [28].

Wearable Devices and the Internet of Things (IoT)

Wearable devices such as smartwatches and fitness trackers can continuously monitor heart rate, sleep patterns, step counts, and activity intensity, providing objective data for lifestyle assessment and behavioral nudges. IoT architectures enable these devices to connect with CGM, smart insulin pens, and home blood pressure or weight monitors, forming an integrated personal health ecosystem. In adults with T2D, wearable technology-based physical activity interventions significantly increase daily step counts and often improve systolic blood pressure, body weight, and other cardiometabolic risk factors, with small but favorable effects on glycemic control [29]. To date, there is a lack of interventional studies specifically evaluating wearable- or IoT-based programs in people with EOT2D, and no robust data on hard clinical endpoints such as hospitalization and cardiovascular-kidney events in this population [29]. While wearables and IoT-enabled devices such as CGM and sensor-equipped monitoring platforms have been studied in T2D populations, there is a notable paucity of research focusing exclusively on people with EOT2D.

Despite this evidence gap, the potential benefits of wearables and IoT for EOT2D are high. People with EOT2D frequently face socioeconomic, work, study, and family pressures that make it challenging to prioritize regular physical activity [29]. However, because this younger population is highly digitally connected, they are uniquely receptive to smartphone‑based support [29]. By capturing physical activity and sleep patterns in real-time, synchronizing these data with glucose and blood pressure readings, and delivering timely feedback, reminders, and goal-setting support, wearable-centric interventions could help people with EOT2D understand their personalized glycemic responses to physical activity, lower the cognitive burden of self-monitoring, and enable more realistic, adaptive physical activity prescriptions. IoT devices may serve as practical tools for empowerment in people with EOT2D, by enhancing self-efficacy, healthy behaviors, and shared decision-making needed to confirm these hypothesized benefits.

mHealth Applications and Digital Therapeutics (DTx)

Smartphone applications currently represent the most prevalent form of DHT in diabetes care. mHealth apps typically combine self-monitoring of blood glucose, dietary, and physical activity with data visualization, personalized reminders, educational modules, goal setting, and in-app communication with healthcare providers, while digital therapeutics (DTx) deliver clinically validated, software-based interventions for disease prevention and management [30]. A 2024 systematic review of 28 digital interventions for adults with T2D (including 23 randomized controlled trials) indicate that structured diabetes management applications, particularly those incorporating personalized feedback, goal setting, and patient-provider communication features, can improve HbA1c, dietary behaviors, and physical activity levels in people with T2D [31].

Direct evidence for mHealth and DTx specifically in people with EOT2D remains limited, as the majority of digital nutrition and lifestyle programs were evaluated in general adults with T2D rather than in those with EOT2D or in head‑to‑head comparisons with usual‑onset T2D. A typical example is the Virta Health remote continuous care model, developed by Virta Health Corp., which integrates app-based low-carbohydrate nutritional therapy, continuous remote biomarker monitoring, and health coaching delivered by a multidisciplinary clinical team [32]. A non-randomized controlled study of adults with T2D was conducted and had successfully achieved a mean HbA1c reduction of 1.3% (from 7.6% to 6.3%) and a mean weight loss of 12% at one year, with sustained improvements observed at two years [33]. However, the study population was not restricted to those with EOT2D, and the lack of age-specific reporting precludes a better understanding of the younger group. Similarly, digital lifestyle programs such as Omada Health, which combine structured online educational curricula, remote health coaching, and peer support, have demonstrated sustained weight loss in large-scale digital diabetes prevention programs [34]. For example, a web-based intervention reported a mean weight loss of 4.7% at 12 months, with durability over longer (up to 2 years) follow-up. However, these studies primarily involved people with prediabetes rather than established EOT2D, and glycemic outcomes were not the primary endpoints [34]. These findings suggest that intensive, coach‑supported digital interventions have strong potential to empower people with EOT2D, by providing flexible, high‑touch support that fits around work and family commitments. Future studies examining their implementation strategies to sustain long-term effectiveness in people with EOT2D are required [35].

Telemedicine and Remote Patient Monitoring (RPM) Platforms

Telemedicine facilitates remote consultations via video, telephone, or asynchronous messaging, offering a way to deliver specialist input without requiring frequent in‑person visits. People with EOT2D often have to juggle multiple commitments, including work and family, on top of requiring frequent clinic visits. Hence, the flexibility of telemedicine and RPM platforms is particularly attractive, as it significantly reduces the physical and psychological challenges to engaging in long-term diabetes care. RPM platforms further enable them to automatically upload data from CGM, smart blood glucose meters, and wearable devices to secure cloud systems, allowing healthcare professionals to remotely review trends and intervene appropriately [36].

Compared with in-person clinical visits, telemedicine care can improve glycaemia in young and middle-aged people with T2D and overweight/obesity, with several remote coaching and RPM programs demonstrating modest HbA1c reductions and improved treatment adherence in younger adults [31, 37–39]. When RPM is integrated with clinical decision support systems, clinicians can rapidly identify people with suboptimal risk factor control or at increased risk of hypoglycemia and those with hypoglycemia unawareness. This enables earlier, more precise treatment adjustments and reduces healthcare utilization, including emergency department visits [40, 41]. In a large digital diabetes RPM program, 63% of participants with poor baseline control reached target glucose or HbA1c on a stable oral and/or injectable regimen, and a 2022 meta-analysis of 20 randomized controlled trials reported that RPM lowered HbA1c by 0.4% over 6 months, supporting earlier, more precise treatment adjustments and better risk factor control [42, 43]. Future studies need to move beyond short‑term feasibility towards hybrid diabetes care models that systematically combine telemedicine, RPM, and in‑person clinic visits. Further evaluations of such models to sustain engagement, improve metabolic and psychosocial outcomes, and narrow access gaps for people with EOT2D are also required [44–46].

Artificial Intelligence and Predictive Modelling

AI algorithms can be applied to the large, high‑frequency datasets generated by CGM, wearables, and mHealth apps to predict short‑term glycemic trajectories, identify high‑risk patterns such as nocturnal hypoglycemia, and provide personalized dietary, physical activity, and treatment recommendations. Multiple reviews and methodological studies show that machine‑learning and deep‑learning models improve predictive accuracy for hypoglycemia, hyperglycemia, and overall glycemic variability compared with traditional approaches, laying the groundwork for data‑driven decision support and ultimately, closed‑loop systems in T2D [47–49]. Adult-onset T2D generally presents with less glycemic variability than T1D. People with EOT2D have an aggressive disease trajectory, rapid pancreatic beta-cell dysfunction, and postprandial glycemic excursions [4]. These are mainly driven by complex behavioral and socio-environmental factors rather than absolute insulin deficiency. The incremental benefit of AI in people with EOT2D extends beyond the optimization of automated insulin delivery (AID) systems. While AID in people with T1D mainly focuses on reactive or short-term predictive insulin dosing, AI applications in those with EOT2D integrate multimodal data, including CGM metrics, dietary intake, physical activity, and medication adherence. Hence, AI in EOT2D shifts the paradigm from algorithmic insulin titration to holistic, preventative risk management that addresses the multifactorial drivers of EOT2D. In addition, among people with EOT2D who face a longer lifetime exposure to hyperglycemia and are at risk of cardiovascular-kidney complications and premature death than those with usual‑onset T2D, such anticipatory tools are particularly relevant because they enable earlier, more precise risk detection and support proactive treatment optimization over decades of disease course [4, 50, 51].

Early clinical and real‑world studies show that AI‑driven personalized nutrition and image recognition-based food‑logging apps can reduce postprandial glycemic excursions, improve glycemic variability, and reduce the burden of manual dietary recording. These can enhance empowerment and self-efficacy in people with T2D [52, 53]. However, most existing models have been developed in small or narrowly defined cohorts and rarely include clearly characterized EOT2D populations. Hence, rigorous external validation across diverse age groups, ethnicities, and sociocultural settings, as well as long‑term evaluation of clinical and psychosocial outcomes in people with EOT2D, should be prioritized [54].

Compared with usual-onset T2D, people with EOT2D tended to have a 0.5-1.0% higher HbA1c at diagnosis, greater obesity and hypertension burden, and a 2–3‑fold higher risk of microvascular and macrovascular complications at any given age [4]. A meta‑analysis of > 1.3 million people reported that, compared with usual-onset T2D, EOT2D was associated with markedly increased risks of cardiovascular disease, kidney disease, and all‑cause mortality, and that each 10‑year earlier age at diagnosis conferred 20–30% higher relative risk of major vascular events and death [55, 56]. Similarly, youth-onset cohorts such as TODAY/TODAY2 showed that by a mean diabetes duration of 13 years (early adulthood), up to 60% of participants develop at least one microvascular complication, and many have multiple complications, underscoring the need for early, intensive, and precisely targeted intervention in this vulnerable group [55, 57–59]. In this context, AI and machine learning–based predictive models, which can outperform traditional scores for predicting incident T2D and its complications, are likely to yield greater lifetime benefit in EOT2D than in usual-onset T2D by enabling earlier identification of high-risk trajectories and more aggressive preventive care over several decades of disease [55, 57, 58].

Key Challenges During Implementation of DHT

The fundamental gap in the current management of EOT2D is the failure of traditional, episodic care models to mitigate the uniquely aggressive trajectory of EOT2D [28, 60]. Traditional care often suffers from high default rates and clinical inertia in this vulnerable group [28]. By leveraging real-time behavioral prompts and gamified digital platforms, DHT provides continuous engagement and empowerment to maintain lifestyle modifications between clinic visits [31, 61]. The sustained behavioral interventions facilitated by DHT could indeed delay the initiation or escalation of glucose-lowering therapies (e.g., insulin, glucagon-like peptide-1 receptor agonists) [32, 33]. This will likely yield cumulative cost savings for healthcare systems, alongside mitigating the burden of premature morbidity [56].

Despite promising prospects, the implementation of DHT in people with EOT2D faces multiple challenges. At the technical level, issues such as poor interoperability between different devices and platforms, a lack of unified data standards, and significant variations in system stability and user experience adversely impact long-term usage and large-scale deployment. These implementation challenges are common to DHT in general rather than being specific to EOT2D alone. Furthermore, insufficient integration with existing healthcare systems, including data security and privacy concerns, means that numerous DHT operate as isolated “information silos”. When these data cannot integrate seamlessly with electronic health record systems, this will not only increase clinicians’ workload but also hinder effective data utilization [62, 63]. Similarly, people with EOT2D live with diabetes across longer and more complex life stages. Fragmented and non‑interoperable systems may disproportionately increase their treatment burden and disrupt continuity of care when they transition from pediatric to adult services.

At the behavioral level, young people from different socioeconomic backgrounds and with varying levels of health literacy exhibit disparities in their acceptance and proficiency with DHT, with the “digital divide” potentially exacerbating existing health inequalities [64, 65]. Simultaneous operation of multiple devices and applications while managing vast amounts of data and reminders may also induce “digital fatigue” and non-adherence to treatment. Sustained long-term use of apps and devices inherently requires motivation and self-discipline. Some studies reported attenuation of intervention effectiveness over time, suggesting the need for design elements such as gamification, personalized notifications, and social support to enhance user retention [31, 61, 66].

At the institutional level, reimbursement policies, regulatory approvals, and liability definitions remain imperfect, constraining the clinical integration and commercial sustainability of innovative tools such as DTx. A few health economics studies reported that many countries still lack stable payment mechanisms and unified evaluation frameworks for DTx, remote monitoring, and digital follow-ups, leaving healthcare institutions facing financial and compliance uncertainties when scaling adoption [67]. Furthermore, legal and ethical norms concerning algorithmic liability, data ownership, and cross-border data flows are evolving, which have also slowed the deep integration of DHT into diabetes care pathways. Figure 2 illustrates how DHT can help address the challenges faced by people with EOT2D.

Future Directions and Research Priorities

Future research and policy efforts should prioritize tailored, life-course–oriented interventions and hybrid care models within broader strategies for managing EOT2D at both individual and population levels [60]. Future DHT of EOT2D needs to transition from ‘tool development’ to ‘system integration’, encompassing evidence-based, person-centered, and seamlessly integrated with existing healthcare systems [68]. The future lies not in complete virtualization, but in the deep integration of online and offline care. Basic monitoring, education, and follow-up can be delivered via digital platforms, while complex diagnostics, treatment planning, and annual complication screening remain the preserve of physical healthcare settings. This model maximizes resource efficiency and the convenience of people with EOT2D [62, 69, 70].

From a research perspective, DHT for EOT2D currently lags behind that for usual‑onset T2D. Future randomized clinical trials and real‑world implementation studies will need to look into people with EOT2D with deliberate assessment of complication incidence, quality of life, mental health, work productivity, and reproductive health, and with subgroup analyses comparing EOT2D and usual‑onset T2D, whenever feasible [71–73].

During the early phases of intervention design, developers should explicitly consider the sociocultural and economic context of people with EOT2D, who are often digitally literate but may have financial instability and fluctuating engagement [55]. Priorities include multilingual content, ultra‑simplified user interfaces that minimize interaction burden, low‑cost or phone‑agnostic solutions, and partnerships with schools, workplaces, and community organizations to deliver digital literacy and peer‑support programs. These features are appealing in Asia and low‑ and middle‑income countries, where the incidence of EOT2D is rising rapidly, but DHT and health inequities remain significant [74].

Finally, precision‑targeted and adaptive interventions that blend behavioral science with AI-driven analytics are needed to move from pilots to sustainable scale‑up, including AI‑supported nutrition tools that are now at the proof‑of‑concept stage [75]. Digital self‑management ecosystems should build on evidence from multifeatured mobile apps and digital DSMES platforms that integrate self‑monitoring, education, coaching, and peer support and have achieved meaningful and, in some cases, HbA1c and weight reductions in adults with long‑standing T2D [76–78]. Flash and continuous glucose monitoring, interoperable data sharing systems, and nurse‑led web‑based education illustrate how integrated monitoring and remote follow‑up can enhance glycemic control, self‑care behaviors, and well‑being when embedded within routine care pathways [79–81]. Digital CBT‑based therapeutics further demonstrate that targeting psychosocial and behavioral drivers can yield additional glycemic benefits beyond standard pharmacotherapy [82]. Multi‑stakeholder collaboration platforms that extend support into home, school, workplace, and community settings, together with reimbursement mechanisms that specifically cover DHT for people with EOT2D, will be essential to adapt these models to younger, high‑risk populations. For additional context, please refer to Supplementary Table 1. Supplementary Fig. 1 illustrates how DHT can be integrated into a layered care model for people with EOT2D.

Human/Animal Studies and Informed Consent Statement

This article does not contain any studies with human or animal subjects performed by any of the authors.

Competing Interests

The author declare the following potential competing interests. Lee-Ling Lim has received honoraria for lectures, presentations, and/or educational activities from Abbott, AstraZeneca, Boehringer Ingelheim, Novo Nordisk, Roche Diabetes Care, and Zuellig Pharma. These payments were made to the individual via her institution and were not related to the preparation of this manuscript. The author declare that these relationships did not influence the content, interpretation, or conclusions of this review. All other authors declare no competing interests.