The Effects of AI-Guided Exercise and a Smart Ring on Arterial Stiffness (GONDOR-AS): protocol for a randomized controlled trial

The GONDOR-AS randomized controlled trial protocol outlines a study comparing the effects of an AI-guided exercise program, supervised high-intensity interval training, and passive monitoring on arterial stiffness and cardiorespiratory fitness in 165 sedentary adults to evaluate the scalability and efficacy of digital health interventions for cardiovascular disease prevention.

The Gist

Imagine your arteries are like old garden hoses. Over time, they can get stiff and rubbery, which makes it harder for your heart to pump blood. This "stiffness" is a major warning sign for heart trouble. On the flip side, your heart's "engine power" (how much oxygen it can use) is like the horsepower of a car. The more horsepower you have, the better your heart works.

This study is like a 12-week fitness experiment designed to see which method is best at turning those stiff garden hoses back into flexible ones and boosting that engine power.

Here is the simple breakdown of what the researchers are doing:

The Three Teams

The researchers recruited 165 adults who usually don't move much. They split them into three different teams, like three different training camps:

1. The "Just Watch" Team (Control Group):

   • The Gear: They get a fancy smart ring (Oura Ring) that tracks their sleep, steps, and heart rate.
   • The Job: They just wear the ring and live their normal lives. They don't get any special instructions to exercise more.
   • The Goal: To see if just knowing you are being watched (and seeing your own data) is enough to make you move more.

2. The "Hardcore Gym" Team (HIIT Group):

   • The Gear: They get the same smart ring.
   • The Job: They go to a lab twice a week to do High-Intensity Interval Training (HIIT). Think of this as sprinting as fast as you can for a few minutes, resting, and repeating. It's supervised by a coach who pushes them to their limit.
   • The Goal: This is the "Gold Standard." We know this works great for the heart, but it's hard to stick with long-term because it's exhausting and requires a coach.

3. The "AI Coach" Team (The Star of the Show):

   • The Gear: They get the smart ring plus a special feature in the app: an AI Coach.
   • The Job: This isn't a human coach; it's a chatbot powered by advanced Artificial Intelligence. It looks at the data from the ring (how you slept, your heart rate, your stress) and chats with you. It says things like, "Hey, you slept poorly last night, maybe take it easy today," or "You're feeling great, let's go for a nice, steady 30-minute walk."
   • The Goal: To see if a friendly, always-available digital coach can get people to exercise enough to fix their arteries, without needing a human trainer or a gym membership.

The Big Question

The researchers want to answer two main questions:

1. Can the AI Coach do as good a job as the Hardcore Gym Team? Can a chatbot get your heart as fit as a grueling, supervised workout?
2. Is the Ring itself enough? Does the "Just Watch" team improve just by wearing the ring, or do they actually need the coaching?

How They Measure Success

They aren't just asking, "Do you feel better?" They are using "Gold Standard" medical tools:

• The Artery Test: They measure how fast a pulse wave travels from your neck to your groin. If your arteries are stiff, the wave travels fast (like a stiff pipe). If they are healthy, it's slower.
• The Engine Test: They have you ride a stationary bike until you can't go anymore to measure your maximum oxygen intake (VO2max).

Why This Matters

We all know we should exercise, but sticking to a routine is the hardest part.

• The Problem: Hiring a personal trainer is expensive and hard to schedule.
• The Hope: If an AI coach on your phone can get you moving just as well as a human trainer, it could be a game-changer. It's like having a personal trainer in your pocket 24/7 who knows your body better than you do, without the high price tag.

The "Secret Sauce" (The AI)

The AI coach is special because it doesn't just give generic advice like "Go run." It builds a memory of you. If you tell it, "I hate running but I love swimming," it suggests swimming. If it sees you slept badly, it suggests a gentle walk instead of a hard run. It creates a feedback loop: You wear the ring -> The ring tells the AI -> The AI talks to you -> You make a change -> The ring sees the change.

The Bottom Line

This study is checking if technology + a friendly chatbot can replace the need for expensive, high-pressure gym sessions to keep our hearts healthy. If it works, it could mean that millions of people could fix their "stiff arteries" just by talking to an app and wearing a ring, making heart health accessible to everyone, not just the gym-goers.

Technical Summary

Based on the provided protocol paper, here is a detailed technical summary of the GONDOR-AS study.

1. Problem Statement

Cardiovascular disease (CVD) remains a leading cause of global mortality, driven significantly by modifiable lifestyle factors. Two critical, modifiable risk factors are arterial stiffness (measured by carotid-femoral pulse wave velocity, cfPWV) and maximal cardiorespiratory fitness (VO2max). While aerobic exercise is proven to improve these metrics, long-term adherence to exercise regimens is a major barrier.
Current digital health solutions often lack the ability to provide continuous, personalized feedback loops that adapt to real-time physiological data. There is a need to determine if AI-guided coaching, integrated with wearable technology, can achieve cardiovascular health benefits comparable to resource-intensive, supervised clinical exercise programs (like High-Intensity Interval Training, or HIIT) while offering a scalable, low-barrier solution for the general population.

2. Methodology

Study Design:

• Type: 12-week, three-arm, parallel-group Randomized Controlled Trial (RCT).
• Participants: 165 sedentary adults (aged 30–65 years).
• Randomization: 1:1:1 allocation into three groups (N=55 per group) using permuted block randomization.
• Blinding: Participants are blinded to their cardiovascular age (CVA) and cfPWV metrics provided by the device; outcome assessors are blinded to group assignment.

Intervention Arms:

1. AI-Based Coaching Group (Intervention): Participants wear an Oura Ring and receive personalized, interactive coaching via a Large Language Model (LLM) integrated into the Oura App. The AI focuses on encouraging moderate, steady-state aerobic exercise ("Zone 2," 2–3 hours/week). The system uses a feedback loop of physiological data (PPG, motion, temperature) and user self-reports to generate non-clinical behavioral guidance.
2. Supervised HIIT Group (Benchmark): Participants wear an Oura Ring and attend twice-weekly, professionally supervised HIIT sessions on cycle ergometers (85% of Wmax4). They also receive a home-based aerobic training plan (150–300 min/week). This group serves as the "gold standard" benchmark for physiological efficacy.
3. Control Group (Passive Monitoring): Participants wear an Oura Ring but receive no exercise intervention or AI coaching. They maintain their usual low activity levels. This isolates the effect of the AI coaching beyond passive self-monitoring.

Key Measurements:

• Primary Outcomes: Change in cfPWV (arterial stiffness) and VO2max (cardiorespiratory fitness) from baseline to 12 weeks.
  • cfPWV: Measured via Complior Analyze (Gold Standard non-invasive device).
  • VO2max: Measured via ergospirometry (breath-by-breath gas exchange) on a cycle ergometer.
• Secondary/Process Measures: Body composition (InBody), physical activity levels (Oura Ring data, IPAQ), exercise self-efficacy (ESES), and qualitative process evaluation (interviews with a subset of 24 participants).

Statistical Analysis:

• Dual Framework: The study employs both Frequentist (Mixed Linear Models, Intention-to-Treat) and Bayesian approaches to provide a robust assessment of evidence.
• Power Calculation: Based on a pilot study, the sample size (55/group) is adjusted for high baseline correlation (R=0.86) and expected attrition, aiming to detect a conservative cfPWV reduction of -0.4 m/s.
• Mediator/Moderator Analysis: Planned to explore how adherence and demographic factors (sex, age, baseline fitness) influence outcomes.

3. Key Contributions

• Novel AI Intervention Architecture: The study utilizes a production-grade LLM-based coaching system ("Oura Advisor") that creates a closed-loop feedback mechanism. Unlike previous chatbot studies, this system integrates real-time, continuous physiological data from a wearable (Oura Ring) to personalize behavioral recommendations dynamically.
• Comparative Effectiveness: It is one of the first RCTs to directly compare a scalable, AI-driven, unsupervised moderate-intensity exercise intervention against a resource-heavy, supervised HIIT benchmark for specific vascular outcomes (cfPWV).
• Process Evaluation Integration: The protocol includes a rigorous mixed-methods process evaluation (interviews, chat log analysis via Nebuly, and questionnaires) to identify the "active ingredients" of the intervention, distinguishing between the effects of the device, the AI, and the exercise itself.
• Methodological Rigor: The use of Gold Standard devices (Complior for PWV, Ergospirometry for VO2max) in a consumer-focused digital health trial sets a high bar for evidence quality in the field of digital therapeutics.

4. Results

• Status: As of the protocol publication (March 2026), the study is ongoing. Recruitment began in October 2024 with a target of 165 participants.
• Current Findings: No primary outcome data is available yet. The protocol notes that data collection is scheduled to conclude in early 2026.
• Preliminary Observations: Early recruitment data suggests excellent adherence, with only three withdrawals during the intervention phase at the time of writing.

5. Significance

• Scalability of Cardiovascular Prevention: If the AI-guided moderate exercise group achieves comparable improvements in arterial stiffness and VO2max to the supervised HIIT group, it would validate AI-driven digital health tools as a viable, scalable alternative to clinical exercise programs for CVD prevention.
• Addressing the Adherence Gap: The study directly addresses the critical challenge of long-term exercise adherence. If successful, it demonstrates that personalized, low-barrier AI coaching can sustain behavior change better than traditional methods.
• Clinical and Commercial Impact: The findings will provide essential evidence for the integration of consumer wearables and AI into preventive healthcare strategies, potentially influencing clinical guidelines and product development in the digital health sector.
• Understanding "Active Ingredients": By comparing the AI group, the HIIT group, and the passive monitoring group, the study will clarify whether the health benefits come from the exercise itself, the motivation provided by the AI, or the mere act of self-monitoring.

Conclusion:
The GONDOR-AS protocol outlines a rigorous investigation into the potential of AI and wearables to revolutionize cardiovascular health management. By leveraging advanced statistical methods and Gold Standard physiological measurements, the study aims to provide definitive evidence on whether automated, personalized coaching can replace or supplement intensive clinical supervision for improving vascular health.