An AI-Powered Lifestyle Intervention vs Human Coaching in the Diabetes Prevention Program: A Randomized Clinical...

Area of Science:

• Digital Health
• Metabolic Health
• Clinical Trials

Background:

• Prediabetes affects a significant portion of the adult population.
• Evidence-based lifestyle interventions, such as the Diabetes Prevention Program (DPP), are underutilized.
• Effective interventions are crucial for preventing progression to type 2 diabetes.

Purpose of the Study:

• To compare the noninferiority of an exclusively artificial intelligence (AI)-led DPP lifestyle intervention versus a human-led DPP.
• To assess outcomes including weight loss, HbA1c reduction, and physical activity levels.
• To evaluate the effectiveness of AI in managing prediabetes in adults with overweight or obesity.

Main Methods:

• A phase 3, pragmatic, noninferiority randomized clinical trial involving adults with prediabetes and overweight/obesity.
• Participants were randomized 1:1 to either an AI-powered DPP (mobile app and digital scale) or a human coach-led DPP.
• Interventions were delivered remotely over 12 months, with outcomes assessed via a composite measure.

Main Results:

• The AI-led DPP group achieved the primary composite outcome in 31.7% of participants, compared to 31.9% in the human-led DPP group.
• The risk difference of -0.2% met the noninferiority criterion, indicating comparable effectiveness.
• High initiation rates were observed for both AI (93.4%) and human-led (82.7%) interventions.

Conclusions:

• Referral to a fully automated AI-led DPP is noninferior to referral to a human-led DPP for adults with prediabetes and overweight or obesity.
• AI-driven interventions show promise as a scalable and effective alternative for lifestyle modification.
• This study supports the integration of AI in managing prediabetes and preventing type 2 diabetes.