Protocol: WorkoutCPP - a pilot series of N-of-1 trials evaluating RL-generated adaptive exercise recommendations for pelvic pain management
Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Introduction
Chronic pelvic pain disorders (CPPDs) affect up to one in four reproductive-age women and are characterized by fluctuating, heterogeneous symptoms that complicate engagement in regular physical exercise. Standard programs rarely adjust to daily symptom variability. WorkoutCPP is a pilot series of N-of-1 trials evaluating the feasibility of an adaptive mobile-health (mHealth) intervention using reinforcement learning (RL) to personalize exercise for individuals with CPPDs.
Methods and Analysis
This pilot study consists of a series of N-of-1 trials employing a randomized ABAB or BABA design, in which 45 participants (aged 18-55 years) with CPPDs each serve as their own control. Each participant completes one baseline week followed by four alternating 2-week intervention phases, totaling 9 weeks, comparing an active comparator arm (A = generic exercise guidance) and an experimental arm (B = adaptive RL-generated recommendations). The RL agent integrated into the ehive app suggests daily exercise recommendations based on participant-reported pain and prior exercise behavior. Primary outcomes are exercise recommendation adherence rate and participant retention rate; safety and additional feasibility indicators are also monitored. Secondary outcomes assess algorithm interpretability. Exploratory outcomes assess contextual sensitivity of the RL algorithm and changes in pain and behavioral engagement.
Ethics and Dissemination
Ethical approval was granted by the Icahn School of Medicine IRB (Protocol # STUDY-23-00721). Electronic informed consent will be obtained. Findings will be shared via peer-reviewed publications, conferences, and de-identified data summaries.
Registration
This study protocol was publicly registered on the Open Science Framework (OSF) Registries (URL: https://osf.io/d9f36 ) prior to participant enrollment, to support international open-science data transparency and reproducibility standards for mobile health adaptive intervention trials. The study was subsequently registered on ClinicalTrials.gov ( NCT07810218 ), consistent with ICMJE recommendations for trials informing future confirmatory research. All documented study procedures match the baseline human subjects tracking mechanisms approved by the Icahn School of Medicine at Mount Sinai Institutional Review Board (Protocol # STUDY-23-00721).
Article Summary
Strengths and Limitations
This study represents the first application of reinforcement-learning–based adaptive exercise recommendations for individuals with chronic pelvic pain disorders.
N-of-1 trial framework enables detailed within-person assessment of feasibility and individual responses.
Integration of wearable and app-based data allows real-time contextual adaptation.
While the ABAB/BABA design supports within-person comparisons, variability across participants (person-level differences) may limit the generalizability of findings.
As a pilot study, the sample is small, findings are exploratory and intended to inform the design of future confirmatory trials.
Although participants are blinded to their assigned arm at any given time, they may infer which study arm they are part of based on the nature of the recommendations.
