The Effects of AI-Guided Exercise and a Smart Ring on Arterial Stiffness: Protocol for a Randomized Controlled Trial
Authors:
- Heikki PentikäinenHPHeikki Pentikäinen
- Sofia StrömmerSSSofia Strömmer
- Drake CarakerDCDrake Caraker
- Julia KosonenJKJulia Kosonen
- Aleksi RantanenARAleksi Rantanen
- Saila HiltunenSHSaila Hiltunen
- Pirjo KomulainenPKPirjo Komulainen
- Heidi SimiläHSHeidi Similä
- Massimiliano de ZambottiMassimiliano de Zambotti
- Kai SavonenKSKai Savonen
- Pauli OhukainenPOPauli Ohukainen
Open Access
- 21 August 2026
- journal article
- Published by JMIR Publications Inc. in JMIR Research Protocols
- Vol. 15, e98213
Abstract
Background: Cardiovascular disease (CVD) prevention is limited by the major challenge of low long-term adherence to effective lifestyle regimens. Arterial stiffness (measured by carotid-femoral pulse wave velocity [cfPWV]) and maximal cardiorespiratory fitness (measured by VO 2 max) are modifiable risk factors for CVD but require sustained lifestyle change. Wearable technology provides continuous measurement and offers a scalable platform to deliver health interventions. A combination of continuous monitoring with a wearable device and an AI-based coach personalized to individual data and preferences could be a powerful, low-barrier tool for achieving sustainable cardiovascular health benefits by directly addressing the adherence challenge. Objective: We will study the comparative effectiveness of a wearable and an interactive app-based AI coaching intervention promoting moderate exercise in improving gold-standard cfPWV and VO 2 max. This will be compared to a supervised high-intensity interval training (HIIT) group (benchmark with known benefits for VO 2 max) and a control group using only the Oura Ring (passive monitoring). We will also conduct a detailed process evaluation (structured interviews) to study the feasibility and experience of interacting with the AI coach. Methods: This randomized controlled trial will recruit 165 eligible sedentary participants aged 30 to 65 years. The co-primary outcomes, cfPWV and VO 2 max, are measured at baseline and will be repeated after 12 weeks. Participants are equally randomized into 3 groups: an AI-based coaching group (steady-state exercise), an HIIT group (supervised exercise), and a control group (usual low activity). The AI-based coaching group receives personalized guidance using large language model technology. All participants wear the Oura Ring and are blinded to the cardiovascular health metrics provided by the ring. Results: The recruitment for the study began in October 2024 and will end when 165 men and women have been recruited. Data collection for the study is scheduled to conclude in early 2026. Data management is ongoing. Conclusions: This study will evaluate whether a potentially highly scalable, AI-based coaching intervention can achieve comparable gains in cardiovascular structural health (cfPWV) and functional capacity VO 2 max relative to a resource-intensive supervised HIIT benchmark. The findings will provide essential evidence on the use of digital health tools to promote sustainable exercise adherence.Keywords
This publication has 22 references indexed in Scilit:
- Global, Regional, and National Burden of Cardiovascular Diseases and Risk Factors in 204 Countries and Territories, 1990-2023JACC, 2025
- Using large language models for narrative analysis: a novel application of generative AIMethods in Psychology, 2025
- Overview of the future impact of wearables and artificial intelligence in healthcare workflows and technologyInternational Journal of Information Management Data Insights, 2024
- Systematic review and meta-analysis of the effectiveness of chatbots on lifestyle behavioursnpj Digital Medicine, 2023
- The effects of aerobic endurance exercise on pulse wave velocity and intima media thickness in adults: A systematic review and meta‐analysisScandinavian Journal of Medicine & Science in Sports, 2015
- The Behavior Change Technique Taxonomy (v1) of 93 Hierarchically Clustered Techniques: Building an International Consensus for the Reporting of Behavior Change InterventionsAnnals of Behavioral Medicine, 2013
- Why We (Usually) Don't Have to Worry About Multiple ComparisonsJournal of Research on Educational Effectiveness, 2012
- A simple sample size formula for analysis of covariance in randomized clinical trialsJournal of Clinical Epidemiology, 2007
- Expert consensus document on arterial stiffness: methodological issues and clinical applicationsEuropean Heart Journal, 2006
- Framework for design and evaluation of complex interventions to improve healthBMJ, 2000