Impact of physical activity on sleep in adults recovering from substance use disorders: a protocol for an N-of-1 observational study Authors: Sascha Thal, Cele Richardson, Joanne McVeigh, Craig Thompson, Phoebe Wan, Bingyan Pang, Hugh Riddell, Stephen Bright, James Clarke, Bronwyn Myers Published: 2025 ABSTRACT ABSTRACTIntroductionSleep impairments are prevalent among individuals recovering from substance use disorders (SUDs) and are associated with poorer treatment outcomes and increased relapse risk. Physical activity (PA) is known to enhance sleep in general populations, but its day-to-day effects on sleep during SUD recovery remain underexplored, especially across different recovery stages.Methods and AnalysisIn this observational n-of-1 study, we aim to examine within-person associations between daily PA and sleep quality in three groups of individuals at varying stages of substance use recovery (early treatment, continuing care and long-term recovery). A substudy will validate the agreement between two wearable sleep-monitoring devices. A total of 90 participants (30 per recovery group) from Western Australia will complete daily ecological momentary assessments and wear the SENS Motion sensor and/or Withings Sleep Analyzer over a 30-day period. Outcomes include device-derived sleep and PA measures, self-reported affect, craving, well-being and recovery outcomes. Multilevel models will explore within-person and between-person associations. This study will generate individualised evidence on associations between PA and sleep to inform tailored SUD care. It will also assess the feasibility of using wearable sleep monitors in real-world SUD recovery settings. Results may support the future development of just-in-time adaptive intervention procedures.Ethics and DisseminationThe study was approved by the Human Research Ethics Committee of Curtin University (HRE2025-0379). Findings will be shared through peer-reviewed publications and conference presentations, with lay summaries provided to treatment facilities and interested participants to support translation into community and service settings.Trial registration numberACTRN12625000835448. STRENGTHS AND LIMITATIONS OF THIS STUDY The study employs an intensive n-of-1 observational design, enabling examination of within-person day-to-day associations between physical activity and sleep across multiple stages of substance use disorder recovery. The study employs an intensive n-of-1 observational design, enabling examination of within-person day-to-day associations between physical activity and sleep across multiple stages of substance use disorder recovery. Use of continuous device-derived monitoring (SENS Motion and Withings Sleep Analyzer) provides objective assessment of sleep and physical activity, reducing reliance on retrospective self-report. Use of continuous device-derived monitoring (SENS Motion and Withings Sleep Analyzer) provides objective assessment of sleep and physical activity, reducing reliance on retrospective self-report. Daily ecological momentary assessment captures fluctuations in affect, craving and well-being, allowing adjustment for temporally varying confounders in the modelling. Daily ecological momentary assessment captures fluctuations in affect, craving and well-being, allowing adjustment for temporally varying confounders in the modelling. Incorporating three distinct recovery groups enables comparison of potential differences in behavioural dynamics across treatment and long-term recovery contexts. Incorporating three distinct recovery groups enables comparison of potential differences in behavioural dynamics across treatment and long-term recovery contexts. Data collection within residential treatment settings may introduce variability in feasibility and completeness of device-derived measurements, particularly for the Withings Sleep Analyzer. Data collection within residential treatment settings may introduce variability in feasibility and completeness of device-derived measurements, particularly for the Withings Sleep Analyzer. INTRODUCTION The Effect of Physical Activity on Sleep in Substance Use Disorder Recovery Substance use disorders (SUDs) represent a significant public health issue, with approximately 2.6% of the global population meeting the criteria for a SUD diagnosis within the past 12 months.1 In Australia, an estimated 2.8%–4.8% of the population meet the criteria for SUD.2 The chronic nature of SUDs,3 high treatment dropout4 and the high prevalence of co-occurring mental and physical health conditions57 add to the complexity and cost of delivering effective care. Extensive research has documented sleep impairments in individuals with SUDs, including alcohol, methamphetamine, nicotine and opioids.810 Insomnia affects approximately 25%–70% of patients during substance use withdrawal,8 11 and sleep disturbances can persist for years after SUD treatment.10 These sleep disturbances are hypothesised to increase the risk of relapse12 and may contribute to the exacerbation of SUDs and the development or worsening of co-occurring mental health conditions like anxiety and depression, as well as chronic pain.9 10 A well-established body of evidence supports the positive effects of sufficient physical activity (PA) on sleep quality, duration and insomnia in various populations, including healthy adults,1316 individuals with Parkinson’s disease,17 depression18 and those at high risk for cardiovascular events.19 While engaging in regular PA can improve multiple aspects of health, such as cardiovascular function, respiratory efficiency, endurance, flexibility, neuromotor abilities and overall physical fitness in individuals with SUDs,20 21 research on the impact of PA on sleep in SUD recovery remains limited. A few recent studies provide preliminary evidence of the positive effects of PA on self-reported sleep quality in clinical populations with non-specified SUD,22 23 heroin use disorders,24 stimulant use disorder25 and former smokers.26 However, these studies relied solely on self-report measures and were limited to participants in the early stages of treatment, without accounting for possible variation in sleep-PA relationships across different points in the recovery process. Importantly, none of these prior studies included device-derived sleep assessments, despite growing evidence suggesting that such measures capture unique and complementary aspects of sleep, and that integrating self-report and device-derived data provides more comprehensive insights into sleep quality than relying on either method alone.2729 Furthermore, previous studies only used between-person designs, which may obscure critical within-person processes that underpin behavioural relationships, challenges and adjustments.30 In contrast, within-person approaches allow researchers to examine how day-to-day fluctuations in an individual’s behaviour relate to changes in their own outcomes: for instance, whether a person sleeps better, feels better and/or experiences less craving on days when they engage in more PA. We respond to these gaps in the literature by examining the daily effects of PA on sleep at different phases of SUD treatment and recovery using device-derived daily PA and sleep measures to capture dynamic, intraindividual processes. Objectives In this study, we aim to explore the relationship between PA and sleep quality over time in individuals undergoing SUD treatment and those in long-term recovery. We also plan to examine the potential relation between sleep quality and SUD recovery outcomes. Additionally, we will conduct a substudy to compare and validate two sleep monitoring devices: the SENS Motion sensor and the Withings Sleep Analyzer. By incorporating repeated device-derived and PA assessments, we seek to provide a more comprehensive understanding of how PA relates to sleep quality both within and between individuals across different stages of recovery. METHODS AND ANALYSIS Study Design We employ an observational parallel n-of-1 design,31 using repeated daily measurements, called ecological momentary assessment (EMA) of sleep, PA and psychological and well-being variables over 30 days within individuals entering SUD treatment (group 1), individuals entering continued care after SUD treatment (group 2), and those in long-term recovery, defined as participants who have been in recovery (ie, abstinence from the primary substance associated with their SUD diagnosis) for 1 year or more (group 3; see figure 1). We followed an extension to the SPIRIT (Standard Protocol Items: Recommendations for Interventional Trials) guideline, the SPENT (SPIRIT extension for n-of-1 trials32) for the development of this protocol (see online supplemental table 1 for SPENT checklist). Participant flow. Note: The figure represents the participant flow throughout the study. EMA, ecological momentary assessment; ORS, Outcome Rating Scale; PANAS, Positive and Negative Affect Schedule. Given the substantial variability in both sleep and PA patterns within and between individuals,10 33 a focus only on group-level analyses may obscure meaningful within-person dynamics. An n-of-1 design addresses this limitation by capturing individual trajectories and day-to-day fluctuations in behaviour that are often masked in aggregated data.34 35 The study uses a parallel N-of-1 framework, in which each participant contributes an individual time series allowing estimation of person-specific associations between daily PA and sleep. This extends traditional multilevel intensive longitudinal designs by enabling not only the pooling of within-person effects at the group level but also the examination of individual trajectories and heterogeneity in dynamic associations. The N-of-1 approach supports the personalisation of findings by allowing individual-level estimates to be inspected, interpreted and compared, thereby identifying patterns that may be clinically meaningful for specific participants. This approach enhances ecological validity. In turn, understanding how PA relates to sleep on an individual level can inform tailored interventions, which align with the principles of precision medicine and individualised treatment and recovery planning.36 37 This is particularly valuable in the context of SUD treatment, where personalised interventions that are responsive to a person’s unique recovery trajectory, level of functioning and readiness for change are considered a critical component of person-centred care and are associated with better SUD treatment outcomes.38 By focusing on within-person evidence, n-of-1 designs yield insights with direct clinical relevance, supporting more adaptive and targeted treatment strategies. Study Setting We will conduct the study across multiple sites in the Perth metropolitan area, Western Australia. Participants will be recruited from three distinct settings: Group 1: residential treatment facilities for individuals undergoing SUD treatment. Group 1: residential treatment facilities for individuals undergoing SUD treatment. Group 2: associated continuing care programme sites that support ongoing SUD recovery following residential treatment. Group 2: associated continuing care programme sites that support ongoing SUD recovery following residential treatment. Group 3: the general community, targeting individuals in long-term (1 year or more) SUD recovery. Group 3: the general community, targeting individuals in long-term (1 year or more) SUD recovery. Eligibility criteria To participate in any of the three study groups, individuals must meet the following criteria: Be aged 18 years or older. Be aged 18 years or older. Have current or past experience with treatment for a SUD. Have current or past experience with treatment for a SUD. Have sufficient proficiency in English to understand study information, provide informed consent and complete study-related questionnaires and tasks independently. Have sufficient proficiency in English to understand study information, provide informed consent and complete study-related questionnaires and tasks independently. Participants will be excluded if they: Have experienced acute relapse in symptoms of a diagnosed comorbid psychotic disorder (eg, schizophrenia) within the past 3 months, and/or they are not stabilised on prescribed medication, as these circumstances may impair capacity to provide informed consent, engagement with daily EMA, or consistent use of wearable devices. Have experienced acute relapse in symptoms of a diagnosed comorbid psychotic disorder (eg, schizophrenia) within the past 3 months, and/or they are not stabilised on prescribed medication, as these circumstances may impair capacity to provide informed consent, engagement with daily EMA, or consistent use of wearable devices. Present with a condition that would substantially interfere with participation in daily assessments or the safe use of wearable devices (eg, significant cognitive impairment, acute medical instability or dermatological conditions preventing sensor adherence). Present with a condition that would substantially interfere with participation in daily assessments or the safe use of wearable devices (eg, significant cognitive impairment, acute medical instability or dermatological conditions preventing sensor adherence). Have a documented physical condition that prevents engagement in any form of PA, as confirmed through participant self-report or clinical information where available. Have a documented physical condition that prevents engagement in any form of PA, as confirmed through participant self-report or clinical information where available. All residents or clients at the partnering residential treatment facilities will be invited to participate. This includes individuals who are either commencing residential treatment (group 1) or transitioning into continuing care (group 2) during the data collection period. As admissions to these facilities occur based on bed availability, participants will enter the study at different time points, resulting in staggered enrolment. Residents who express interest in participating will be screened for eligibility by the research team. Individuals in long-term recovery from the general community (group 3) will also be invited to participate following expression of interest and subsequent eligibility screening. Screening Procedures Screening will be conducted by trained research staff following expression of interest. For individuals in groups 1 and 2, initial eligibility will be discussed with staff at the participating treatment facilities to confirm suitability based on their clinical information and current treatment status. Eligibility for all groups will then be assessed using a structured screening checklist that includes SUD treatment history, the study’s exclusion criteria, and the individual’s capacity to engage with the EMA procedures and wearable devices. For some group 3 participants who are referred or who independently approach the study in response to distributed materials, treatment history and duration of recovery will be based on self-report. Measures and Instruments Sample Characteristics Demographics Participants will be asked to provide demographic details such as their age, gender, country of birth, employment status, regular medication use and treatment history. Typical PA We will use the Global Physical Activity Questionnaire (GAPQO)39 to gather data on participants’ PA behaviour prior to their SUD treatment. Participants will report on their usual daily PA across three settings or domains: work activities, travel to and from places, and recreational activities, based on a typical 7-day period within the past year. They will also answer a general question about the time spent being sedentary each day during the same period. The GAPQ consists of 16 items and has been well validated.40 Sleep Quality Index The Pittsburgh Sleep Quality Index (PSQI41) will be used to assess participants’ sleep quality over the past month. The PSQI captures self-reported information across seven components: subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of sleep medication and daytime dysfunction. These components are combined to generate a global score ranging from 0 to 21, with higher scores indicating poorer sleep quality. The instrument consists of 19 items and has demonstrated strong reliability and validity in both clinical and non-clinical populations.42 43 Substance Use History Participants will be asked to choose from a list the substances they have used regularly (ie, on a weekly basis) prior to starting treatment. A similar questionnaire has been used in past studies.4446 Treatment Completion For group 1 participants who consent to their treatment data being shared, treatment centres will provide the date on which the participant either completes the residential programme or exits prior to completion. Treatment completion status will be used descriptively and to support planned post hoc comparisons between participants who complete treatment and those who exit early. Ecological Momentary Assessment Device-derived PA The PA variables analysed will include daily step count, sedentary time and active minutes, as measured by SENS motion devices. The SENS motion system includes a lightweight, waterproof sensor equipped with a triaxial accelerometer that samples acceleration at 12.5 Hz within a±4G range. The sensor connects via Bluetooth to a dedicated app, and participants will be asked to synchronise the sensors once a week to upload the data. It is embedded in a Medipore soft cloth surgical tape and applied to the skin on the outer thigh, about 10 cm above the lateral knee (lateral epicondyle). Active minutes are defined as the time spent in various intensities of PA and postures (ie, sitting, standing, walking, running, etc). The device is a reliable and validated measurement system for PA and sedentary behaviour.47 Device-derived Sleep Measure We will assess sleep using Withings Sleep Analyzers and SENS motion. The Withings Sleep Analyzer48 features a thermoplastic polyurethane pad filled with air and linked to a pressure sensor. It is placed under the mattress, beneath the patient’s torso. In essence, the sensor detects pressure changes within the air bladder relative to atmospheric pressure. Additionally, the device includes a microphone, also positioned under the mattress. The collected pressure and sound data are processed using embedded software. Total sleep time is determined by the total number of minutes the user is detected in bed and classified as asleep. Sleep efficiency is calculated as total sleep time divided by the total minutes the user is present in bed. The algorithm further detects apnoea and hypopnea events during sleep. The Withings Sleep Analyzer was shown to reliably detect sleep apnoea48 and showed accurate sleep-wake distinction.49 The SENS motion collects data on estimated sleep and wake times, as well as movement during sleep. Using the ‘Activity and Sleep’ algorithm, the devices provide an assessment of sleep quality based on metrics such as nocturnal movement and sleeping position. The algorithm is designed to detect the main overnight sleep period, identifying sleep onset between 18:00 and 1:00, and wake time between 5:00 and 12:00. It focuses solely on nighttime sleep and does not capture or classify daytime naps as sleep episodes.50 SENS Motion and Withings Sleep Analyzer Validation Study We will embed a substudy within the main study to compare and validate two sleep monitoring devices: SENS motion and the Withings Sleep Analyzer. This substudy was designed based on practical considerations. While all participants will wear SENS motion devices to track their PA and sleep, we anticipate that not all participants in Group 1 will be able to use the Withings Sleep Analyzer within their treatment facility. Additionally, facility staff may not be able to support data collection from the device in accordance with the study protocol (see below). Therefore, use of the Withings Sleep Analyzer will be optional for participants in group 1. In contrast, all participants in groups 2 and 3 will use the Withings Sleep Analyzer alongside the SENS motion device to monitor both sleep and PA (see table 1). Table 1Set-up of substudyGroup n Physical activity measurementSleep measurementGroup 1: Participants starting treatment30SENSSENSSENS+WithingsGroup 2: Participants in continued care30SENSSENS+WithingsGroup 3: Participants ≥1 year recovery30SENSSENS+Withings Set-up of substudy As not all participants in group 1 will be able to use the Withings device, between-group comparisons will primarily rely on sleep metrics derived from the SENS Motion device, which is used consistently across all groups. The rationale for this dual-device strategy is to maximise sleep measurement feasibility across diverse settings while enabling an exploratory assessment of cross-device agreement. Withings-derived sleep data will be used only where agreement between devices is supported by the validation sub-study and may be explored in post-hoc comparisons between groups 2 and 3. Self-reported Sleep Quality The consensus sleep diary51 will require participants to self-record their sleep patterns, including sleep timing, sleep onset latency, wake after sleep onset and total sleep time. In addition, sleep quality will be self-reported on a 7-point Likert scale, indicating how rested participants felt: 0 (very poor), 1 (poor), 2 (a bit poor), 3 (adequate), 4 (a bit rested), 5 (rested) and 6 (very rested), and how they rate their sleep quality: 0 (very poor), 1 (poor), 2 (fairly poor), 3 (neutral/neither good nor poor), 4 (fairly good), 5 (good) and 6 (excellent). Similar methods have been employed in previous EMA studies that used self-reports to evaluate sleep quality.52 Sleep diaries have been recognised as a reliable tool for assessing sleep-wake patterns.53 Self-reported Affect Participants will answer six questions regarding their positive and negative emotions over the past day. These questions have been shown to be effective for assessing emotional states through EMA in both depression-prone and control groups.54 Each question begins with: “During the past day, how strongly did you feel…?” The emotions included are four negative ones (sad/blue, nervous/anxious, irritable and angry) and two positive ones (happy, satisfied), taken from the Positive and Negative Affect Schedule.55 Participants will be asked to indicate on a visual scale how strongly they felt the respective emotions from 0 (not at all) to 100 (very much). For the subscale analysis, the mean scores for positive and negative affect will be calculated by averaging the relevant items. Self-reported Recovery Outcomes The four-item Outcome Rating Scale (ORS)56 will be used to assess daily changes in psychosocial functioning. The ORS is a brief self-report measure designed to capture a client’s subjective well-being across key life domains, which may be indicative of the overall treatment outcome. It asks participants to reflect on their functioning over the past day and rate themselves on four Visual Analogue Scales: (1) individual or personal well-being (eg, physical health, emotional health), (2) interpersonal relationships (eg, family, close relationships), (3) social functioning (eg, work, school, friendships) and (4) overall sense of well-being. Scores on each scale range from 0 (low functioning) to 100 (high functioning), with a maximum total score of 400. Higher scores indicate greater subjective well-being and functioning. The ORS has been validated for use in clinical and community settings, demonstrating sensitivity to change over time and strong concurrent validity with longer outcome measures.56 Self-reported Craving The three-item Craving Scale57 will be used to measure daily substance use cravings. The three items assess: (1) the strength of the participant’s self-reported desire to use alcohol and/or other drugs in the past 24 hours (0=no desire to 100=extremely strong); (2) the likelihood of use if the participant were in a previous use-related environment (0=not at all to 100=I am sure I would use) and (3) the strength of urges when encountering environmental reminders of use (0=no urges to 100=extremely strong). The total craving score is obtained by averaging the three items. This scale has been well-validated across various SUDs.58 Outcomes Primary Outcome The primary outcome is daily sleep quality, assessed via device-derived sleep indicators (SENS Motion for all groups, with supplementary Withings Sleep Analyzer data where available). The core metrics include sleep duration, sleep efficiency and sleep–wake patterns captured across the 30-day monitoring period. Secondary Outcomes Secondary outcomes include: Daily PA, measured via SENS Motion (step counts, activity intensity, sedentary time). Daily PA, measured via SENS Motion (step counts, activity intensity, sedentary time). Self-reported daily affect, craving and well-being, captured through EMA surveys. Self-reported daily affect, craving and well-being, captured through EMA surveys. Covariates All other daily and baseline variables measured will be included as covariates in relevant models, depending on analytic requirements. Adverse Events and Unintended Effects Although the EMA protocol does not capture adverse events directly, treatment facilities will record any clinically relevant events occurring among participants in groups 1 and 2 and will forward this information to the research team where appropriate. For group 3, adverse events cannot be systematically monitored during the study period, but any incidents reported by participants during the device return and debrief session will be documented. Any unintended effects arising from study conduct (eg, discomfort with daily assessments or device use) will be documented if reported by participants, noted in study records and, where necessary, communicated to the Human Research Ethics Committee in accordance with institutional requirements. Sample Size N-of-1 Observational Study Within-Person Analysis To ensure that within-person estimates are based on a sufficient number of observations, we will include participants in individual-level time-series analyses only if they provide at least 14 days of daily data. Simulation-based work on within-person reliability indicates that very small numbers of occasions per person yield unstable within-person estimates, with acceptable reliability typically reached only once roughly 10 or more observations are available for many realistic combinations of intraclass correlation coefficient (ICC), scale length and sample size.59 Similarly, simulation studies of daily cyclic processes suggest that at least 10 measurements per participant are required even to detect large effects, with smaller effects requiring substantially more repeated measurements.60 On this basis, a 14-day minimum represents a conservative threshold for obtaining reasonably stable individual-level estimates in the present intensive longitudinal design. Between-Person Analysis Power simulations were conducted using the simr package in R,61 employing a linear mixed-effects model (LMM) with 90 participants each contributing 30 repeated observations. Assuming a moderate within-person effect of PA on sleep quality (β=0.31), with random intercepts and slopes by individual, the simulation (1000 iterations) demonstrated a power of approximately 100% to detect the fixed effect at a significance level of α=0.05. Based on simulation studies by Arend and Schäfer,62 a sample with 90 participants (level 2) and 30 repeated daily measurements per participant (level 1) is sufficient to detect small within-person effects and moderate between-person effects with adequate power. Specifically, for level 1 predictors, they report a minimum detectable standardised effect size of approximately 0.19 to achieve ≥80% power (α=0.05), while for level 2 predictors, a standardised effect size of 0.30 is sufficient under the same conditions. These results confirm that the current study design is well-powered to detect meaningful associations between PA and sleep at both the within-person and between-person levels. SENS Motion and Withings Sleep Analyzer Validation Study For the validation component, the aim is to assess the agreement between the Withings Sleep Analyzer and the SENS Motion system by comparing estimates of total sleep time. The primary statistical approach will involve calculating the ICC for continuous sleep parameters, with the ICC serving as the primary metric of agreement. Each participant is expected to contribute approximately 30 nights of sleep data, enabling the use of a repeated-measures design to enhance statistical power. We assume an expected ICC of 0.80 (good agreement) and a minimum acceptable ICC of 0.60 under the null hypothesis. Using established formulas63 64 for ICC power analysis in the context of repeated measures, and assuming a significance level of α=0.05 and 80% power, a minimum of 10–12 participants is estimated to be sufficient to detect the expected level of agreement. As the study aims to recruit 30 participants per group, we anticipate obtaining data for both devices (groups 2 and 3) from 60 participants for the validation component. This sample size is expected to provide more than adequate statistical power for assessing agreement between the two devices. Participant Recruitment Participants will be recruited on a rolling basis until complete datasets are obtained for 30 individuals in each recovery group (N=90 in total). Given the anticipated higher dropout rate among those beginning treatment (group 1), additional recruitment efforts will be directed toward this group. We will leverage existing partnerships with treatment facilities in Perth (Tenacious House and Cyrenian House) and relationships with individuals who have lived experience with SUD treatment, established through previous research projects to facilitate recruitment. Individuals newly admitted to the residential treatment facility (group 1) or to a continuing care programme (group 2) will be provided with an introduction to the study and a copy of the participant information sheet by facility staff as part of their intake process. Before any data is collected, members of the research team will deliver a clear and concise verbal explanation of the study. Residents will then be invited to take part. Printed copies of the information sheet and consent form will be handed out, and participants will have the opportunity to ask questions and discuss the study with the researchers before deciding whether to participate. Eligibility criteria are clearly explained in the participant information sheets. Researchers will review these criteria with individuals who express interest and assess their eligibility accordingly. On providing informed consent, participants will be asked to complete a brief survey covering demographic details, sleep quality, usual PA and substance use history. The survey can be completed either online via Qualtrics (using an iPad provided by the research team or their personal mobile device) or on a paper-based version, under the supervision of a researcher. In addition, participants will complete the EMA survey to provide baseline measures of affect, well-being and craving. This also offers an opportunity for them to ask questions if they are unsure about any of the survey items. These surveys are expected to take approximately 10–15 min to complete. For group 3, recruitment will proceed through two structured pathways. First, we will engage the partnering treatment facilities collaborating in groups 1 and 2 (Tenacious House and Cyrenian House), who will contact eligible past clients to inform them of the study and invite them to express interest directly to the research team. Participants who enrol through this pathway will also be invited to refer peers in long-term recovery, forming a controlled snowball sampling stream with clearly defined starting points. Second, we will distribute study information through established community-facing recovery organisations and peer-led groups with whom we have ongoing collaborations, including 12-step groups and Cyrenian House outpatient services. Study materials will be disseminated via service staff at these organisations. Interested individuals will be invited to meet with a member of the research team at a mutually agreed on, comfortable and private location (eg, a quiet café, community centre or the university). During this meeting, researchers will provide a clear explanation of the study, along with printed copies of the participant information sheet and consent form. Individuals will be encouraged to ask questions and will be given the opportunity to consider their involvement before providing informed consent. As with other groups, consenting participants will then complete the same brief surveys. Approximately 30 sleep analysers and 30 activity monitors will be available for the study. Recruitment started on 9 September 2025 and data collection is expected to take around 6−7 months and is anticipated to finish on 31 March 2026. As a thank you for their time and involvement, participants will be invited to enter a prize draw on completing the study. Three participants will be randomly selected to each receive a US$100 voucher for activewear of their choice. Entry into the draw is voluntary and not a condition of participation. The draw will be conducted at the conclusion of data collection, and winners will be notified via their preferred contact method. This remuneration model is consistent with Australian ethical guidance.65 Patient and Public Involvement One author identifies as someone with lived experience of SUD recovery and was involved in shaping elements of the study design and implementation. This contributor had previously guided codesign processes in an earlier project46 66 that informed the development of several components of the present protocol. They were further engaged to support participant recruitment and provide additional feedback on feasibility considerations in real-world recovery settings once data collections started. Their lived experience contributed to decisions regarding the suitability of the daily EMA schedule, the acceptability of the wearable devices and the likely burden of daily participation across different recovery settings. Study findings will be disseminated to participants and relevant services through accessible summaries and presentations, with input from the contributor on content and format to ensure they are useful and meaningful to individuals in recovery. Data Collection Methods After completing the baseline survey, the data collection process will vary across study groups. For group 1, each participant will be provided with a SENS motion device (the use of the Withings Sleep Analyzer is optional, and its feasibility will be assessed during participant recruitment), and an iPad, on which individual EMA surveys will be created using Qualtrics. EMA surveys to assess daily emotions, cravings and sleep will be completed each evening using iPads and under the supervision of facility staff. Due to restrictions on the use of multimedia devices in the facility, the iPads will remain offline and will only be distributed for the purpose of completing daily surveys. The Qualtrics app, the Withings Sleep Analyzer app, and the SENS motion app will be preinstalled on the iPads, with the activity monitors and sleep analysers connected to the iPads via Bluetooth. In the evenings, facility staff will distribute the iPads among participants and prompt them to complete the survey, subsequently collecting and securely storing the devices in a locked filing cabinet. At that time, the SENS motion and the Withings Sleep Analyzer will be synced with the iPad through the Bluetooth connection. The iPads, connected to the internet, then upload the EMA survey responses and the PA and sleep data daily (see figure 2). For groups 2 and 3, a similar setup will be used; however, participants’ personal smartphones will replace the iPads (and participants will receive daily messages including survey links), with PA and sleep data synced directly to apps on their devices. Participants’ accounts for the respective apps will be set up by the researchers as part of the enrolment procedure. Participants in group 1 will be instructed to complete the EMA survey at 18:00 each evening, and participants in groups 2 and 3 will receive automated survey prompts at the same time on their personal devices. The survey platform automatically records the exact completion time, and this timestamp will be included as a covariate in the analyses to account for potential diurnal variation. All participants will receive a debrief including a personalised summary of their SENS motion data, accompanied by explanations from the research team, at the conclusion of their data collection. Study setup. Note: Each participant was provided with a SENS Motion device attached to their thigh and (optionally) a Withings Sleep Analyzer that was put under their mattress. They completed daily questionnaires either on iPads (group 1) or their phones (groups 2 and 3). Individual EMA surveys were created using Qualtrics. The Qualtrics app, the SENS Motion app and the Withings app were preinstalled on the iPads or downloaded on the participants’ phones, with the SENS Motion and the Withings Sleep Analyzer connected to the iPads or phones via Bluetooth. Participants completed the EMA surveys on the iPads or phones each evening. At that time, the SENS Motion app and the Withings app were opened on the devices to automatically sync their activity and sleep data through the Bluetooth connection. The iPads or phones, connected to the internet, then automatically uploaded both the EMA survey responses and the physical activity and sleep data daily. Dotted arrows in the figure represent the connections between devices, while solid black arrows illustrate the flow of data from devices to the cloud. EMA, ecological momentary assessment; PA, physical activity. Data Management Hard copy data will be securely stored in locked filing cabinets on university premises, with no personal identifiers included. Any digital files containing identifiable participant information will be kept on the University’s secure server, protected by password access. Only the research team will have permission to access both physical and digital data. Comprehensive procedures for data handling and storage are outlined in a data management plan, which has been reviewed and approved by the University’s Human Research Ethics Committee. A formal Data Monitoring Committee is not required for this study. The protocol uses an observational n-of-1 design without experimental interventions or clinical endpoints, and participant risk is minimal. No interim analyses or stopping guidelines are planned, as there is no intervention to terminate; oversight is provided through approved ethics procedures, secure data management and review by the University’s Human Research Ethics Committee, making an independent DMC unnecessary. Data Analysis N-of-1 Observational Study This study generates intensive longitudinal data from parallel N-of-1 time series, in which each participant contributes their own within-person dataset. Analyses will be conducted at both the individual and group levels. For individual-level analyses, we will model each participant’s daily associations between PA and sleep using person-specific time series estimates; these results will be presented descriptively for participants with adequate data. At the group level, multilevel models will pool the within-person parameters across participants to estimate average effects while accounting for between-person heterogeneity. We will report descriptive statistics for demographic characteristics, typical PA, sleep indexes and substance use for each group and the full sample. Each participant’s data will be treated as a distinct dataset and analysed individually. Missing data will be handled using multiple imputation with bootstrapping, implemented via the Amelia II package in R.67 We will conduct imputation separately for each participant and each variable (ie, one time series per question per participant) to preserve individual temporal dynamics and avoid borrowing information across participants. Time-indexed variables (eg, linear time trends and lagged values where available) will be included as predictors to maintain the temporal structure inherent in intensive longitudinal data, as recommended for N-of-1 designs. Baseline variables will not be imputed. Multiply imputed datasets will be analysed separately, with estimates pooled according to Rubin’s rules. To evaluate the robustness of the findings, results from imputed datasets will be compared with complete-case analyses. Potential differences in EMA adherence between staff-supported surveys (group 1) and mobile phone surveys (groups 2 and 3) will be examined descriptively, and any systematic discrepancies across groups will be explored through sensitivity analyses. As a first step, daily trajectories of PA (step counts, active minutes in various intensities), sleep metrics (eg, total sleep time, sleep efficiency, wake after sleep onset) and psychological variables (mood and craving scores) will be plotted for each participant across the study period and visually inspected for trends, patterns and anomalies. We will employ LMMs68 69 for the core analysis to investigate the within-person association between PA and sleep quality. LMMs are appropriate for this design as they account for the non-independence of repeated measures within individuals and allow for individual variation in baseline levels and trajectories. Model diagnostics will be performed prior to interpretation, including assessment of linearity, normality of residuals, homoscedasticity and detection of influential outliers via residual plots, Q-Q plots and Cook’s distance. If assumptions are violated, transformations (eg, log, square root) or robust modelling techniques will be considered. The models will include: Fixed effects for daily PA metrics as predictors. Fixed effects for daily PA metrics as predictors. Relevant covariates such as prior night’s sleep quality, psychological states and craving intensity. Relevant covariates such as prior night’s sleep quality, psychological states and craving intensity. Random intercepts and slopes for PA by participant to capture individual differences in baseline sleep and PA-sleep associations. Random intercepts and slopes for PA by participant to capture individual differences in baseline sleep and PA-sleep associations. An autocorrelation structure or lag terms to address temporal dependencies in repeated measures. An autocorrelation structure or lag terms to address temporal dependencies in repeated measures. Additionally, time-lagged analyses will explore whether daily PA predicts sleep on the following night, adjusting for previous sleep quality and psychological variables, to investigate potential directional effects. Group-level comparisons (eg, between participants in treatment, continuing care or long-term recovery) will be conducted by including interaction terms for group membership. Interaction terms will be fitted between group membership (groups 1, 2, 3) and the primary daily predictors (eg, daily PA, previous-night sleep, affective states), allowing us to examine whether within-person associations differ by recovery context. We will use a multilevel location-scale model in a dynamic structural equation modelling framework70 to explore study variables as predictors of within-person variability as well as mean levels of sleep. All analyses will be conducted using Mplus, with α set at 0.05 and 95% CIs reported. Bayesian multilevel models may be used as an alternative estimation framework to a maximum likelihood estimator because they accommodate sparse within-person time series, provide regularised individual-level estimates and offer greater flexibility for estimating complex temporal structures. Their use is therefore based on methodological suitability for intensive longitudinal N-of-1 data, as well as their improved computational efficiency and precision,71 rather than solely on model convergence considerations. Post hoc and exploratory analyses In addition to the primary analyses specified above, we plan to conduct some post-hoc exploratory analyses where appropriate. These analyses will focus on patterns that may be relevant for interpreting variability in sleep and PA across different recovery contexts. As an initial planned comparison, we will examine sleep-related trajectories among participants recruited from residential treatment, comparing those who complete the programme with those who exit prior to completion. Further exploratory analyses may include: (1) evaluating whether PA performed at different times of day shows differential associations with subsequent sleep outcomes; (2) analysing nocturnal restlessness and sleep-related movement as distinct components rather than as aggregated sleep indices; (3) exploring whether changes in affect or sleep predict early disengagement from the study or treatment and (4) examining the extent to which self-reported sleep measures agree with device-derived indicators. Additional post hoc analyses may be added where justified by emerging data patterns and feasibility considerations. SENS Motion and Withings Sleep Analyzer Validation Study The validation study will assess the agreement between the Withings Sleep Analyzer and the SENS Motion system for key sleep parameters, including total sleep time and sleep efficiency. We will calculate descriptive statistics for these parameters for each device separately. The primary analysis will involve quantifying agreement between devices using the ICC for continuous sleep metrics. ICC values will be interpreted according to established benchmarks, with values above 0.75 indicating good agreement and values above 0.90 indicating excellent agreement.72 Bland-Altman plots will be generated to visually examine the mean differences and limits of agreement between devices across the range of measurements. In addition, paired t-tests or equivalent non-parametric tests will be conducted to test for statistically significant differences in mean sleep parameters between devices. Data quality and consistency will be assessed by inspecting measurement distributions, missing data patterns and any outlier values. Sensitivity analyses will be conducted by excluding nights with incomplete or questionable data to evaluate the robustness of findings. All analyses will be conducted in R, using irr73 and BlandAltmanLeh74 packages. Protocol Amendments We will record any changes to the study protocol and share them when the study results are published. Confidentiality The information collected in this study will be reidentifiable only by the research team. Any information collected will be confidential and only used for the purpose of the study. All data will be stored securely and in adherence with National Statement on Ethical Conduct in Human Research75 and the Western Australian University Sector Disposal Authority.76 ETHICS AND DISSEMINATION The study was approved by the Human Research Ethics Committee of an Australian University (HRE2025-0379) and was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12625000835448) on 4 August 2025, prior to recruitment of the first participant. The findings from this study will be disseminated through peer-reviewed journal publications and presentations at national and international conferences. In line with open science principles, an anonymised dataset will be deposited on the Open Science Framework following completion of the project, consistent with the data curation and retention procedures outlined in the approved data management plan. Lay summaries of the study outcomes will also be provided to participating treatment facilities and shared with interested participants to support the translation of findings into community and service contexts. DISCUSSION Our study aims to address a notable gap in the literature concerning the relationship between PA and sleep during SUD recovery. Sleep disturbances are often persistent and long-lasting throughout the recovery process and have been identified as a critical factor that may undermine the effectiveness of PA-based interventions in this population.9 10 12 By investigating potential associations between PA and sleep, sleep and affect, and sleep and recovery, this research has the potential to help inform and tailor SUD treatments to the unique needs of individuals. Limitations This study protocol has several methodological limitations that arise primarily from the need to minimise participant burden and operate within available resources. First, this study does not track medication use, withdrawal severity, pain, daytime fatigue or other clinical events longitudinally, which may contribute to unmeasured confounding in the associations between PA and sleep. Environmental differences between residential treatment and community settings may also introduce contextual confounding that cannot be fully controlled analytically. These limitations reflect the feasibility constraints of the present study and will be acknowledged when interpreting findings. Second, the demographic information collected at intake is necessarily restricted. Biological sex was not recorded due to ethical guidance to prioritise self-identified gender as the least distressing option for participants in treatment and recovery. This decision limits the capacity to examine sex-based differences in PA, sleep and recovery-related outcomes. Third, detailed substance use histories, including duration and quantity of use, were not collected to reduce respondent burden during an already extensive intake process. The absence of this information may constrain the ability to account for substance-specific chronicity or dosage-related associations with sleep, PA and recovery trajectories. Fourth, the EMA surveys do not include free-text fields for qualitative comments as, in our experience, this substantially increases survey completion time. However, this may limit contextual understanding of participants’ experiences. Fifth, for participants in group 3, treatment history and duration of recovery will be based on self-report for those recruited through community channels or referrals, which may introduce recall bias. Recruitment pathways for this group may also lead to selection and network-related sampling biases, despite efforts to engage multiple community-facing services and peer-led recovery groups. Sixth, as the Withings device may only be used by a subset of group 1 participants, agreement estimates may be influenced by self-selection. Strengths and Contributions The findings from this study may contribute to the development of more targeted interventions that address both PA and sleep-related challenges in individuals recovering from SUDs. By providing a clearer understanding of how these factors interact over time within individuals, the study could support the refinement of existing treatment programmes to better promote long-term recovery and overall health. In addition, the results may guide the design and implementation of new interventions specifically aimed at improving both PA and sleep. If scaled effectively, such programmes have the potential to not only enhance recovery outcomes but also reduce the elevated risk of chronic conditions commonly observed in this population.5 Ultimately, this could lead to improved quality of life for individuals in recovery and reduced long-term healthcare costs. An important methodological contribution of this study lies in demonstrating the feasibility of collecting continuous, device-derived sleep data in populations undergoing SUD treatment and recovery. Individuals in these settings often experience instability, fluctuating engagement and varying levels of functioning, which can pose significant challenges to sustained data collection using wearable or contactless monitoring devices. By implementing the use of the SENS Motion sensor and the Withings Sleep Analyzer across diverse treatment and recovery contexts, this study may produce evidence on the acceptability and feasibility of device-derived sleep monitoring within this population. These findings are crucial for informing the design of future trials that aim to evaluate the efficacy of sleep-targeted or PA-targeted interventions in real-world SUD treatment settings. Moreover, the integration of wearable technology into longitudinal observational research strengthens the validity of sleep assessments and reduces reliance on self-report measures, which are often limited by recall bias and subjective interpretation, particularly in populations affected by disrupted sleep patterns. Establishing the practicality and value of such methods represents a critical step towards more rigorous and personalised intervention research in SUD recovery. Eventually, the integration of these just-in-time adaptive intervention methods77 78 may allow health professionals to provide dynamic behavioural support in response to real-time sleep changes to improve treatment and health outcomes. This direction is echoed in the UK National Health Service’s 10-Year Plan,79 which highlights the routine integration of wearable technologies as a cornerstone of preventative and chronic disease management, reinforcing the urgency and relevance of incorporating such tools into SUD treatment research to future-proof care systems and support scalable, personalised health interventions. Supplementary material