Study: AI-Led Diabetes Prevention as Effective as Human Coaching

Spread the Message

Read Time:4 Minute, 46 Second

BALTIMORE — A fully automated artificial intelligence (AI) program is just as effective as human coaches in helping people with prediabetes reduce their risk of developing type 2 diabetes, according to a landmark study led by researchers at Johns Hopkins Medicine.

The findings, published recently in the Journal of the American Medical Association (JAMA), suggest that AI-driven interventions could solve a critical public health challenge: how to scale effective diabetes prevention to millions of people amid a global shortage of healthcare professionals.

The Breakthrough Finding

The Phase III randomized clinical trial compared a “human-led” Diabetes Prevention Program (DPP)—the current gold standard—against an AI-powered app that used algorithms to provide personalized coaching.

The results showed a statistical “tie.” After 12 months, 31.7% of participants using the AI app met the Centers for Disease Control and Prevention (CDC) benchmarks for reducing diabetes risk. This was statistically non-inferior to the 31.9% success rate seen in the human-coached group.

“Even beyond diabetes prevention research, there have been very few randomized controlled trials that directly compare AI-based, patient-directed interventions to traditional human standards of care,” said Nestoras Mathioudakis, M.D., M.H.S., co-medical director of the Johns Hopkins Medicine Diabetes Prevention & Education Program and the study’s lead investigator.

Higher Engagement with AI

While the clinical outcomes were similar, the study revealed a distinct advantage for the AI group regarding accessibility and engagement.

Because the AI program was available on-demand via a smartphone app, it eliminated common barriers such as scheduling conflicts or the need to travel to a clinic. Consequently, the AI group saw significantly higher participation rates:

Program Initiation: 93.4% of AI participants started the program, compared to 82.7% in the human-led group.
Completion Rate: 63.9% of AI users completed the study, versus only 50.3% of those assigned to human coaches.

“Unlike human-coached programs, AI-DPPs can be fully automated and always available, extending their reach and making them resistant to factors that may limit access to human DPPs, like staffing shortages,” noted study co-author Dr. Benjamin Lalani.

Study Details and Methodology

The trial enrolled 368 adults with prediabetes and a body mass index (BMI) indicating overweight or obesity. The median age was 58. Participants were randomly assigned to one of two groups for a 12-month period:

AI Group: Received a smartphone app and a Bluetooth-enabled scale. The app used reinforcement learning algorithms to send personalized “push notifications” and coaching tips based on the user’s weight, diet, and physical activity data.
Human Group: Received a standard, CDC-recognized prevention program led by human coaches, delivered remotely (telehealth).

The “success” of the treatment was measured by a composite score defined by the CDC. To meet the benchmark, a patient had to achieve at least 5% weight loss, or a combination of 4% weight loss and 150 minutes of weekly physical activity, or a meaningful reduction in hemoglobin A1c levels (a marker of blood sugar control).

Implications for Public Health

Prediabetes affects nearly 98 million American adults, yet only a small fraction participate in the National Diabetes Prevention Program. A major bottleneck has been the limited availability of trained human coaches and the high cost of delivering labor-intensive programs.

The validation of an AI-led model suggests that healthcare systems could deploy these tools to vastly more patients at a lower cost.

“While the ‘black-box’ nature of AI is a commonly cited barrier to clinical adoption, our study shows that the AI-DPP can provide reliable personalized interventions,” Lalani added.

Independent observers note that while AI lacks the “ineffable” human element of empathy, its consistency and convenience may make it a more practical option for the majority of patients who struggle to fit traditional medical appointments into their lives.

Limitations and Future Context

The study authors acknowledged certain limitations. The trial was conducted during the COVID-19 pandemic, which may have influenced participants’ willingness to engage with digital tools versus in-person care. Additionally, the study excluded patients who were already taking diabetes medications like metformin or GLP-1 agonists (e.g., Ozempic), meaning the results apply specifically to lifestyle-based prevention.

Furthermore, while the AI matched human coaching in efficacy, it did not surpass it. For some patients, the accountability provided by a human relationship remains a powerful motivator that an algorithm cannot fully replicate.

As the medical community continues to integrate digital health tools, this study provides the rigorous evidence needed to prescribe an “app” with the same confidence as a human referral.

Medical Disclaimer: This article is for informational purposes only and should not be considered medical advice. Always consult with qualified healthcare professionals before making any health-related decisions or changes to your treatment plan. The information presented here is based on current research and expert opinions, which may evolve as new evidence emerges.

References

Primary Study: Mathioudakis, N., Lalani, B., et al. “An AI-Powered Lifestyle Intervention vs Human Coaching in the Diabetes Prevention Program: A Randomized Clinical Trial.” JAMA. Published October 27, 2025. DOI: 10.1001/jama.2025.19563.
Institutional Source: Johns Hopkins Medicine Newsroom. “AI-Powered Diabetes Prevention Program Shows Similar Benefits to Those Led by People.”
Statistics: Centers for Disease Control and Prevention (CDC). National Diabetes Statistics Report.
Independent Commentary: Shaywitz, D. “The Outsized Significance of A New Study of AI in Diabetes Prevention.” Timmerman Report, Oct 30, 2025.

This video provides an overview of the Johns Hopkins Diabetes Prevention and Education Program, offering context on the human-led interventions that were compared against AI in the study.