For the roughly one-third of adults worldwide living with at least one chronic condition, the gap between what happens inside a clinic and what happens in daily life has long been a blind spot in care. A comprehensive review now maps exactly where digital tools — wearables and AI — are genuinely closing that gap, and where the hype still outpaces the evidence.
The review, published in Frontiers in Public Health, synthesizes findings across six major chronic disease categories: type 2 diabetes, obesity, cardiovascular disease, chronic respiratory disease, cancer survivorship, and multimorbid older adults. Wearable devices — tracking steps, activity intensity, sedentary time, heart rate, and sleep — showed the most consistent evidence of benefit for behavioral outcomes: measurable increases in daily step counts, reduced sedentary time, and improved self-monitoring adherence. Evidence for functional improvements such as exercise capacity and intermediate clinical markers (HbA1c, blood pressure, weight) was described as promising but heterogeneous across studies. Long-term clinical endpoints, cost-effectiveness data, and scalable health-system integration remain poorly characterized. AI-specific contributions — including personalized feedback loops, risk stratification, and dynamic goal adjustment — were flagged as largely feasibility-stage, with limited rigorous trial evidence supporting their clinical value independent of the wearable platform itself.
This review arrives at a pivotal moment when health systems in many countries are being pressured to adopt digital health infrastructure at speed. Its core message — that wearables and AI represent components of a broader "digital public health closed loop" rather than standalone clinical interventions — is a valuable counterweight to vendor-driven enthusiasm. The finding that behavioral outcomes are the most robustly improved is significant but also sobering: moving more is necessary but not sufficient to alter hard endpoints like cardiovascular mortality. A key limitation is that narrative and scoping reviews of this type aggregate heterogeneous study designs, making effect-size comparisons inherently imprecise. The field urgently needs adequately powered randomized trials with standardized outcome hierarchies before AI-enabled physical activity tools can be recommended as evidence-based clinical tools rather than promising adjuncts.
