A longitudinal investigation tracking stroke survivors during their community recovery phase demonstrates that real-time physical movement is heavily governed by immediate environmental and social circumstances, with an individual’s psychological self-control playing a pivotal buffering role against acute psychological strain. Researchers from Fudan University and Soochow University examined ambulatory behavior using an intensive ecological momentary assessment framework, collecting continuous objective biometric data alongside repeated daily context surveys. The investigation revealed that post-stroke walking volume and intensity remain substantially insufficient across daily life, yet fluctuate markedly depending on where patients are, what activities they perform, and whom they accompany. By identifying how physical location, occupational commitments, and interpersonal engagement interact with personal self-regulatory capacity, the study provides an empirical, mechanism-based foundation for designing context-sensitive rehabilitation regimens tailored to individuals in their natural living environments.
Regular walking activities can improve clinical outcomes in stroke patients via multiple pathways. Previous clinical literature underscores this therapeutic necessity, noting that stroke survivors who achieve more than 8,000 steps per day exhibit an annual reduction of 0.03 millimeters in carotid intima-media thickness progression, representing an estimated 27 to 31 percent reduction in relative cardiovascular risk. Furthermore, achieving 6,000 daily steps has been shown to reduce recurring cardiovascular events following mild ischemic stroke. Despite these established clinical advantages, stroke survivors typically achieve only half the daily step count observed in age-matched healthy peers. Prior observational research over a two-year post-discharge period documented that barely one-quarter of stroke patients maintained more than 6,500 steps per day at six, twelve, and twenty-four months, underscoring an enduring challenge in sustaining physical rehabilitation outside formal clinical facilities.
Traditional investigations into post-stroke physical activity have predominantly relied on retrospective self-report questionnaires administered during periodic clinic follow-ups. While informative regarding broad population averages, these retrospective assessments are notoriously susceptible to recall bias and fundamentally fail to capture short-term behavioral fluctuations that unfold across dynamic daily contexts. Moreover, conventional rehabilitation studies have concentrated primarily on isolated main effects of environmental barriers or static psychological traits, leaving the dynamic interplay between immediate environmental demands and internal self-regulatory resources largely unexamined. To overcome these limitations, the research team implemented an ecological momentary assessment protocol grounded in person-environment interaction theory, which posits that human behavior emerges from continuous transactions between personal traits and situational context, providing the precise empirical data needed for developing just-in-time adaptive digital interventions.
The study protocol recruited adult stroke patients from the neurology outpatient and inpatient records of a grade III class A hospital in Suzhou, Jiangsu Province, China, between January and July 2025. Enrolled participants were required to be at least eighteen years of age, possess confirmed definitive stroke diagnoses, exhibit intact verbal communication and cognitive abilities, maintain access to smartphones with WeChat proficiency, and ambulate independently without physical assistance. Crucially, participants had to be at least three months post-hospital discharge and maintain a modified Rankin Scale score of 2 or less, denoting mild or absent functional disability. Potential subjects presenting severe cardiovascular, pulmonary, or active psychiatric comorbidities, as well as those with serious complications such as heart failure, were excluded to prevent confounding ambulatory measurements.
From an initial enrollment cohort of 154 eligible stroke patients, 130 participants completed the full monitoring protocol, reflecting a study completion rate of 84.4 percent and a dropout rate of 15.6 percent. Among the 24 non-completers, eight participants cited excessive burden from wearing monitoring hardware, four withdrew due to personal reasons, four exhibited protocol non-compliance, and eight generated fewer than three valid days of sensor data. The final analytical sample of 130 patients was predominantly male (68.5 percent) and married (98.5 percent), with 52.3 percent aged 65 years or older and 47.7 percent under 65 years. The vast majority of patients had experienced ischemic stroke due to cerebral arterial thrombosis (98.5 percent), while 1.5 percent presented ischemic stroke complicated by hemorrhagic transformation. Furthermore, 99.2 percent recorded a Barthel Index score of 60 or higher, confirming fundamental independence in essential daily living activities.
Objective ambulatory monitoring was accomplished using wrist-worn Huawei HA580 smart wristbands fitted to each patient’s dominant wrist for seven consecutive days. These wristbands incorporated precision accelerometers and gyroscopes capable of transmitting continuous step count, cadences, and physiological variables directly to a centralized server. Walking volume was operationalized as cumulative step count, while walking intensity was defined as the cumulative duration of moderate-to-high-intensity activity, determined using an empirical cadence threshold of at least 100 steps per minute. A preliminary validation sub-study conducted among 21 stroke inpatients performing three-minute walking trials against video-recorded criterion standards demonstrated a device measurement error rate of 9.19 percent, well within accepted biomechanical monitoring tolerances. Valid patient-days required at least 10 hours of documented daily device wear time, determined via optical sensor heart rate signals, with an observed mean wear time of 21.35 hours per day.
To evaluate psychological attributes and momentary ecological circumstances, researchers combined baseline psychometric assessments with five scheduled daily electronic questionnaires. At baseline, individual self-control was quantified using the seven-item Brief Self-Control Scale, which measures impulse suppression and self-discipline on a 35-point scale, demonstrating strong internal consistency with a Cronbach’s alpha of 0.808. Baseline self-regulatory fatigue was evaluated via the sixteen-item Self-Regulatory Fatigue Scale across cognitive, emotional, and behavioral control domains, yielding an alpha coefficient of 0.84. Throughout the active monitoring week, each calendar day was divided into five distinct intervals: early morning (04:00 to 07:59), forenoon (08:00 to 11:59), afternoon (12:00 to 15:59), evening (16:00 to 19:59), and night (20:00 to 02:59). At the conclusion of each window, participants completed surveys recording physical location (indoor versus outdoor) and social context (solitary, social or recreational, or working status), achieving an overall questionnaire compliance rate of 94.5 percent.
In parallel with behavioral self-reports, real-time psychological stress was evaluated using heart rate variability indices generated continuously by the wristband’s integrated optical sensor array and proprietary AI algorithmic monitoring technology. Because hierarchical clustering is inherent in longitudinal data, the research team analyzed the resulting 4,550 participant-day-time-segment observations using three-level generalized linear mixed models. Level 1 captured repeated momentary observations within specific time segments, Level 2 accounted for day-to-day variance within individual participants across the 910 recorded person-days, and Level 3 accounted for stable between-person differences. Highly skewed step counts were fitted using negative binomial generalized linear mixed models, whereas moderate-to-vigorous physical activity durations were modeled using zero-inflated negative binomial formulations to accommodate excessive zero-minute intervals, with continuous contextual indicators centered around group means and individual traits centered around grand means.
The collected activity data revealed marked deficits and pronounced variability in ambulatory behavior. Across the 910 evaluated participant-days, the median daily step count was 7,713, spanning a minimum of 246 steps to a maximum of 23,122 steps. Only 24.7 percent of participant-days met or exceeded the public health threshold of 10,000 steps per day, while 43.1 percent fell below 5,000 daily steps. Activity intensity proved even more constrained, with participants registering a median cumulative duration of walking at 100 steps per minute or higher of just 8 minutes per day. A striking 85.7 percent of monitored person-days failed to achieve the widely recommended guideline of 30 minutes of moderate-intensity ambulatory movement, demonstrating that even when stroke survivors accumulate moderate total step volumes, high-cadence cardiovascular training remains critically underdeveloped.
Multilevel fixed-effects modeling demonstrated that situational factors exerted profound direct influences on momentary physical activity. When compared against early morning baseline segments, step counts were significantly elevated during the forenoon (incident rate ratio [IRR]: 1.50), afternoon (IRR: 1.35), and evening (IRR: 1.28), while plunging during nighttime periods (IRR: 0.57). Physical location proved crucial, as outdoor environments increased step counts by 67 percent compared to indoor locations (IRR: 1.67). Similarly, engaging in active work duties (IRR: 1.79) or participating in social interactions (IRR: 2.49) produced substantial step count surges compared to solitary states. Parallel trends emerged for moderate-to-vigorous physical activity duration, which increased markedly outdoors (IRR: 2.30), in social settings (IRR: 3.88), and during work (IRR: 2.57), while falling under elevated maximum daily temperatures (IRR: 0.98) and elevated momentary physiological stress (p = 0.027).
Importantly, the multi-level analysis confirmed that baseline personal traits significantly moderated the impact of immediate situations, directly validating person-situation interaction principles. In Model 4, personal self-control significantly moderated the relationship between employment status and total step volume (IRR: 0.940, 95% CI: 0.889 to 0.994). Conditional effect decomposition showed that among stroke patients exhibiting low self-control (-1 standard deviation), being at work generated an exceptionally large boost in walking volume (IRR: 2.08) compared to resting at home.
A complementary moderating effect of self-control was uncovered when evaluating the restrictive impact of acute psychological strain on higher-intensity activity. In Model 4, self-control significantly moderated the relationship between momentary physiological stress and moderate-to-vigorous physical activity duration (IRR: 1.004, 95% CI: 1.001 to 1.007). Conditional analysis established that among patients with low self-control (-1 standard deviation), acute stress significantly suppressed higher-intensity walking duration (IRR: 0.983, 95% CI: 0.973 to 0.993). In patients with moderate self-control, this negative association persisted (IRR: 0.991), but among patients possessing high self-control (+1 standard deviation), the inhibitory effect of acute stress disappeared entirely (IRR: 1.000). Conversely, baseline self-regulatory fatigue exhibited no statistically significant moderating effects across situational variables, indicating that general fatigue scores do not alter situational sensitivity in the presence of dominant physical impairments.
The study authors acknowledged important methodological boundaries, including convenience recruitment from a single regional hospital setting in Suzhou, which circumscribes demographic generalizability to broader stroke populations. Furthermore, the device-derived stress index was generated based on the wearable device algorithm and was not independently validated against established psychological stress assessment methods in the study population.
Subject of Research: Person-situation interactions and ambulatory activity dynamics in post-stroke rehabilitation
Article Title: Individual-situation interactions influencing walking activity in post-stroke patients: an ecological momentary assessment study
Article References: Pan, X., Chen, L., Wang, L., Zhang, Y., Chen, Y., Xu, L., & Yuan, C. (2026). Individual-situation interactions influencing walking activity in post-stroke patients: an ecological momentary assessment study. Frontiers in Digital Health, 8, Article 1909918. https://doi.org/10.3389/fdgth.2026.1909918
Image Credits: AI Generated
DOI: 10.3389/fdgth.2026.1909918
Keywords: stroke recovery, ecological momentary assessment, walking activity, step count, moderate-to-vigorous physical activity, generalized linear mixed models, wearable accelerometers, heart rate variability, self-control, self-regulatory fatigue, person-situation interaction, social environment, outdoor physical activity, cadence threshold, rehabilitation medicine, just-in-time adaptive interventions, stroke rehabilitation
Tags: cadence thresholdecological momentary assessmentgeneralized linear mixed modelsheart rate variabilityjust-in-time adaptive interventionsmoderate-to-vigorous physical activityoutdoor physical activityperson-situation interactionrehabilitation medicineself-controlself-regulatory fatiguesocial environmentstep countstroke recoverystroke rehabilitationwalking activitywearable accelerometers
