News Research Glucose Monitoring & Insulin Delivery Ethics, Regulation, and Responsible Use Remote Monitoring and Wearables

AI in pediatric diabetes: Benefits and risks

October 26, 2026 By Julie Greenbaum 7 min read
Share Share via Email Share on Facebook Share on LinkedIn Share on Twitter

Artificial intelligence-enabled technologies, particularly automated insulin delivery systems, improve glycemic outcomes in children with type 1 diabetes, but the long-term developmental, psychosocial, and ethical effects of exposure to both medical and consumer artificial intelligence remain insufficiently understood.

Researchers conducted a purposive narrative literature search for a review examining artificial intelligence (AI) in pediatric diabetes, published in the European Journal of Pediatrics, using PubMed/MEDLINE, Scopus, and Google Scholar and covering publications from database inception through February 2026.

The review focused primarily on type 1 diabetes, incorporating evidence related to type 2 diabetes when relevant. Ultimately, 19 publications were selected for the core narrative synthesis, with additional references providing background and supporting information.

Automated insulin delivery improves glycemic outcomes

The strongest clinical evidence identified in the review supported automated insulin delivery systems, which integrate continuous glucose monitoring data with algorithms that adjust insulin administration to maintain glucose concentrations within a target range.

A recent meta-analysis involving pediatric populations reported an approximately 11% increase in time in range and a reduction in glycated hemoglobin (HbA1c) of approximately 0.4% percentage points, without an increase in severe hypoglycemia or diabetic ketoacidosis.

Beyond glycemic outcomes, the review identified evidence that automated diabetes technologies may reduce fear of hypoglycemia, diabetes-related distress, and caregiver burden. Continuous glucose monitoring and predictive algorithms may also improve parental sleep, reduce stress, and support shared decision-making.

Chatbots and digital support show emerging benefits

Emerging evidence suggests that AI-based conversational agents and digital coaching platforms may support diabetes self-management, although pediatric-specific evidence remains limited.

The review identified studies suggesting that diabetes chatbots may improve patient engagement, disease-related knowledge, self-efficacy, and selected glycemic outcomes. The researchers cited the My Diabetes Care application as an example demonstrating the feasibility of conversational support for diabetes self-management.

Robotic educational platforms incorporating large language models have also shown potential to improve diabetes knowledge and achieve high acceptance among children.

However, sustained engagement remains a challenge. The researchers described evidence of declining participation following initial adoption of digital interventions. They cited findings from the BlueStar platform, in which highly engaged users experienced glycemic benefits, although randomized trials did not demonstrate significant HbA1c reductions across all participants amid inconsistent engagement.

The review also identified dissatisfaction with chatbot interactions that users perceived as repetitive, scripted, or emotionally limited. The researchers noted that maintaining long-term engagement with these technologies may introduce additional diabetes-management demands.

Continuous monitoring raises developmental questions

Children and adolescents with diabetes may experience increased digital interaction through diabetes applications, glucose monitoring, alerts, and AI-enabled support systems. However, whether medically necessary screen exposure has developmental consequences comparable to recreational screen use remains unknown.

Continuous glucose monitoring offers clinical benefits but may also introduce psychosocial challenges. Although remote monitoring can enhance safety by alerting caregivers to impending hypoglycemia, frequent alerts may contribute to parental anxiety, sleep disruption, and heightened vigilance.

The researchers described the concept of the “quantified child,” in which continuous collection of physiological data may increasingly shape childhood experiences. They noted that excessive parental monitoring could conflict with adolescent privacy and autonomy, particularly when glucose information influences restrictions on social activities.

The review did not establish that continuous monitoring causes adverse developmental outcomes. The threshold at which beneficial monitoring may become excessive surveillance remains unknown.

Generative AI introduces additional concerns

Although generative AI may provide personalized education and immediate access to diabetes-related information, these systems can generate inaccurate medical advice, present misleading information with confidence, and create false perceptions of expertise. The researchers noted that children may struggle to distinguish between evidence-based recommendations and algorithmically generated misinformation.

AI companions may offer accessible, nonjudgmental opportunities for emotional expression, particularly for adolescents reluctant to discuss diabetes-related concerns with caregivers or health care professionals.

The researchers noted that frequent interactions with AI companions could influence patterns of social engagement in some users. However, current evidence does not demonstrate that AI companions replace human relationships, and longitudinal research is needed to determine whether these technologies supplement or displace interpersonal interactions.

Privacy, consent, and unequal access

The review identified unresolved questions regarding ownership and governance of the extensive biometric data generated by diabetes technologies.

The researchers noted that many families remain unaware of how pediatric diabetes data are stored, shared, or used for future algorithmic training. They also emphasized that existing frameworks provide limited guidance regarding pediatric consent for AI-driven interventions, particularly as children develop greater autonomy in health care decision-making.

Access to advanced diabetes technologies represents another concern. The review described associations between technology adoption and socioeconomic status, insurance coverage, educational attainment, and geographic location.

The researchers also identified the potential for algorithms trained predominantly on data from populations with greater access to technology to perform less effectively in underserved populations.

Whether AI ultimately reduces or increases disparities in pediatric diabetes care remains unresolved.

Proposed approach to pediatric diabetes care

To address these competing benefits and concerns, the researchers proposed shifting the emphasis of AI-assisted care from “quantifying the child” toward “supporting the child and caregiver.”

Their conceptual framework prioritizes reducing caregiver cognitive burden, improving family communication, protecting privacy, and preserving children's developing autonomy.

The review also recommended maintaining physician oversight of algorithm-driven care and incorporating AI literacy into diabetes education. Suggested strategies included reviewing the accuracy of glucose monitoring systems, individualizing alert settings, discussing data privacy, and encouraging families to verify AI-generated medical information with their diabetes care team before making treatment decisions.

The researchers emphasized that the proposed framework is conceptual and has not been established as an evidence-based clinical model.

Study limitations

The researchers acknowledged several limitations, including the narrative review design, heterogeneity of AI technologies, and limited longitudinal evidence examining developmental, psychosocial, and ethical outcomes among children with diabetes.

Some concerns discussed represent emerging hypotheses rather than established causal relationships. The rapidly evolving nature of AI also means that its clinical applications and evidence base will continue to change.

How these findings affect patient care

Hussein Zaitoon, MD, of the Division of Pediatric Endocrinology and Diabetes at the University of Texas Health Science Center at San Antonio and an author of the review, emphasized in comments provided for this article that AI should complement, rather than replace, diabetes education, clinical judgment, and family-centered care.

Hussein Zaitoon, MD, of the Division ofPediatric Endocrinology and Diabetes at the University of Texas Health Science Center at San Antonio
Hussein Zaitoon, MD

“AI literacy should be part of diabetes care education. Clinicians may increasingly need to discuss how children and families use AI-generated health information, particularly when generative AI is used for diabetes questions or self-management decisions,” he said. “Discussions should also address appropriate use of diabetes technology, privacy, alert burden, digital well-being, and the importance of maintaining age-appropriate independence and family communication.”

“Artificial intelligence is increasingly becoming part of pediatric diabetes care, with the strongest evidence supporting automated insulin delivery and other algorithm-driven technologies that improve glycemic control and reduce diabetes management burden,” said Dr. Zaitoon. “At the same time, clinicians should recognize that children are exposed to AI beyond the clinical setting, including generative AI and other consumer technologies. The goal should be to use AI to support children and families while preserving clinical judgment, human relationships, and the child's developing autonomy.”

Cautions and next steps for research

The clinical benefits of AI-enabled diabetes technologies are increasingly well established, but much less is known about the long-term developmental, psychosocial, and ethical effects of sustained AI exposure during childhood, Dr. Zaitoon explained.

“Future research should move beyond glycemic outcomes to evaluate child development, autonomy, family functioning, psychological well-being, privacy, health equity, and long-term self-management,” he said. “Longitudinal studies will be particularly important in determining how AI-assisted diabetes care affects children as they progress through adolescence and into adulthood.”

The researchers declared no competing interests. No funding was reported.

AACE Endocrine AI is published by Conexiant under a license arrangement with the American Association of Clinical Endocrinology, Inc. (AACE®). The ideas and opinions expressed in AACE Endocrine AI do not necessarily reflect those of Conexiant or AACE. For more information, see Policies.

Related Content