Effectiveness Of Artificial Intelligence-Driven Lifestyle Interventions For The Prevention And Management Of Type 2 Diabetes: A Systematic Review With Narrative Synthesis

Abstract

Background: Type 2 diabetes mellitus (T2DM) has become a global epidemic in recent decades, contributing substantially to morbidity, mortality, and economic burden. The increasing prevalence of T2DM is largely attributable to ageing populations, urbanization, unhealthy diets, inadequate physical activity, and rising obesity rates. While lifestyle modification is a primary strategy for T2DM prevention and management, long-term behavioural modification is challenged by poor adherence, limited access to healthcare, and suboptimal counselling in personalised interventions. Objective: This systematic review with narrative synthesis aimed to evaluate the impact of artificial intelligence-based lifestyle interventions on glycaemic, anthropometric, lifestyle, and self-management outcomes among individuals with T2DM, prediabetes, or at high risk of T2DM. Methods: This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 guidelines. Literature was searched in PubMed/MEDLINE, Embase, Scopus, Web of Science, Cochrane CENTRAL, IEEE Xplore, CINAHL, ProQuest, ClinicalTrials.gov, and the WHO International Clinical Trials Registry Platform (ICTRP) for studies published from 1 January 2020 to 31 December 2025. Eligible studies included adults with prediabetes, at high risk of developing T2DM, or with T2DM who received an AI-enabled lifestyle intervention. Interventions included AI-enabled mobile applications, machine learning algorithms, conversational chatbots, virtual health coaches, digital therapeutics, predictive analytics, and wearable-based AI platforms. Randomised controlled trials or controlled clinical studies reporting glycaemic, anthropometric, behavioural, or self-management outcomes were eligible. Results: The literature search yielded 1,186 studies, of which seven RCTs involving 2,369 participants met the inclusion criteria and were included in the narrative synthesis. The included studies originated from the United States, Australia, Spain, Czech Republic, Sweden, and Singapore. AI technologies included mHealth applications, conversational AI, digital behavioural therapies, AI coaching systems, and AI-assisted wearable technology, with intervention periods ranging from 3 to 12 months. Overall, AI-supported lifestyle interventions showed positive effects on glycaemic control, with all but one study reporting significant reductions in HbA1c. Positive effects were also reported for body weight, body mass index, physical activity, dietary adherence, medication adherence, patient engagement, and diabetes self-care activities. Conclusion: AI-supported lifestyle interventions show considerable promise as an adjunct to standard diabetes management through individualised recommendations, real-time monitoring, and ongoing behavioural guidance. These systems may help optimise diabetes prevention and management, particularly in resource-constrained healthcare settings. Further well-designed multicentre RCTs with standardised AI methods, longer follow-up periods, and economic evaluations are warranted.