Effects of Artificial Intelligence Recognition-Based Telerehabilitation on Exercise Capacity in Patients With Hypertension: Randomized Controlled Trial

Abstract

Background: Hypertension remains a major global health challenge, significantly increasing cardiovascular and all-cause mortality risks. While exercise therapy is effective, conventional approaches face limitations in accessibility and personalization, compromising adherence. Artificial intelligence (AI)-assisted remote rehabilitation enables real-time monitoring and personalized guidance, offering a promising alternative. Nevertheless, its clinical benefits and applicability require further systematic validation.

Objective: This study aimed to evaluate the efficacy of an 8-week AI-assisted telerehabilitation program on improving exercise capacity and related health outcomes in patients with hypertension.

Methods: This prospective, dual-arm, parallel, open-label, randomized controlled trial enrolled 62 patients with hypertension recruited via convenience sampling. Participants were adults aged between 18 and 75 years with a confirmed hypertension diagnosis who were excluded for severe cardiac complications, recent myocardial infarction, unstable angina, or physical disabilities preventing exercise. The participants were randomly assigned (1:1) to an intervention group that received AI-assisted remote rehabilitation plus routine health education, or a control group that received health education and conventional offline exercise guidance. The supervised exercise program included warm-up, cardiorespiratory endurance, strength resistance, balance, and flexibility training, followed by a cooldown. Sessions lasted between 30 and 50 minutes and were performed at least 3 times weekly for 8 weeks. Assessments at baseline and 8 weeks included the 6-minute walk test (6MWT), cardiopulmonary exercise testing (CPET), International Physical Activity Questionnaire (IPAQ), Short-Form Health Survey 12 (SF-12), Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7), exercise self-efficacy, blood pressure (BP), body weight, handgrip strength, and other health-related indicators. The primary outcome was the change in 6-minute walk distance (6MWD). Data were analyzed according to the intention-to-treat principle.

Results: Throughout the 8-week intervention period, no serious adverse events related to the AI-assisted telerehabilitation intervention occurred. After 8 weeks, the intervention group demonstrated significantly greater improvements than the control group in 6-minute walk distance (6MWD; adjusted mean difference 62.77, 95% CI 26.33-99.22; P=.002), systolic BP reduction (adjusted mean difference 4.11, 95% CI 0.11-8.28; P=.046), IPAQ score (adjusted mean difference 658.96, 95% CI 159.23-1158.69; P=.011), exercise self-efficacy score (adjusted mean difference 21.71, 95% CI 13.59-29.82; P<.001), total exercise time (adjusted mean difference 98.24, 95% CI 49.39-147.08; P=.001) peak oxygen uptake (peak VO₂) (adjusted mean difference 3.39, 95% CI 0.49-6.29; P=.026), and peak oxygen uptake percent predicted (peak VO₂%pred) (adjusted mean difference 11.58, 95% CI 2.06-21.10; P=.021).

Conclusions: Compared with conventional exercise rehabilitation, AI-assisted remote rehabilitation was found to improve exercise capacity, boost regular physical activity and exercise self-efficacy, and aid in systolic BP control among patients with hypertension. This study positioned AI-assisted rehabilitation as a scalable and effective strategy for real-world hypertension management. It further contributes actionable guidance for developing effective home-based exercise strategies tailored to populations with hypertension.

Trial registration: Chinese Clinical Trial Registry ChiCTR2300076451; https://www.chictr.org.cn/showproj.html?proj=208353.

Keywords: artificial intelligence; exercise habit formation; hypertension; lifestyle change; randomized controlled trial; telerehabilitation.