Stroke is the second-leading cause of death and the third-leading cause of disability worldwide, imposing a substantial burden on society and families [1]. Advances in acute stroke management have improved survival rates; however, most survivors experience persistent functional impairments [2]. In addition to motor, sensory, swallowing, and speech impairments, stroke frequently induces respiratory dysfunction during the acute stage [3]. Recent studies indicate that 18% to 88% of patients with stroke exhibited abnormal breathing patterns [4], while 44% to 90% present with respiratory failure syndrome of varying severity [5]. Furthermore, 7% to 38% develop pneumonia within the first week after the onset of acute stroke [6], underscoring the urgent need for effective interventions to improve respiratory function.

Clinical guidelines advocate early respiratory monitoring and intervention to mitigate poststroke respiratory complications, emphasizing an integrated care that includes physical and respiratory rehabilitation [7]. Previous studies have demonstrated that respiratory muscle training and breathing exercises, such as diaphragmatic, air-stacking, and pursed-lip breathing, can improve respiratory muscle strength, pulmonary function, and functional capacity during early stroke rehabilitation [8-12]. However, the implementation of respiratory interventions in stroke rehabilitation remains limited by multiple clinical constraints, including insufficient staffing and inconsistent service delivery [13-16]. In many health care settings, rehabilitation services are delayed or underused due to the prioritization of acute medical management, leaving survivors with stroke with inadequate support for respiratory dysfunction [17-19]. In this context, telerehabilitation may represent a promising alternative for delivering timely and accessible respiratory rehabilitation interventions.

Telerehabilitation has emerged globally as a promising approach to improving access to and continuity of stroke rehabilitation services, attracting widespread attention [20,21]. Chen et al [22] reported that telerehabilitation has been widely applied in stroke rehabilitation and provided effects comparable to conventional rehabilitation in improving motor function among survivors with stroke. Linder et al [23] demonstrated that mobile-based exercise interventions can effectively enhance quality of life and alleviate depressive symptoms among individuals recovering from stroke. In addition, Sun et al  [24] found that telerehabilitation can effectively reduce caregiver burden. Moreover, a growing body of evidence from high-income countries supports the feasibility, safety, and effectiveness of telerehabilitation in the postacute and chronic phases of stroke recovery [25-27]. However, despite these encouraging findings, the application of telerehabilitation during the acute hospitalization phase of stroke remains largely unexplored.

Therefore, this study aimed to evaluate the effectiveness of a 2-week hospital-based comprehensive mobile-based respiratory training program (CMRTP) on respiratory function in patients with acute stroke, and to explore its safety and feasibility for early inpatient rehabilitation. We hypothesized that integrating CMRTP with conventional rehabilitation would yield greater improvements in respiratory function, without increasing the incidence of adverse events.

A total of 56 patients were screened for eligibility, and 40 were enrolled in the study. Of these, 20 were randomly assigned to the intervention group and 20 to the control group (Figure 4). One participant in the intervention group withdrew during the second week due to hospital discharge. Regarding treatment adherence, participants in the experimental group completed 192 of the 200 prescribed training sessions (96%), whereas those in the control group completed all 100 assigned sessions (100%). Outcome data for analysis were available for 100% (N=40) of participants at T1 and 97.5% (39/40) at T2. Nonserious adverse events considered reasonably or definitely related to study procedures were recorded in 10 sessions in the experimental group (including fatigue, dizziness, and hyperventilation) and in 3 sessions in the control group (including fatigue and tiredness). All events were transient and mild, and no serious adverse events occurred in either group.

The baseline clinical characteristics of the patients, shown in Table 1, were generally comparable between the experimental and control groups. All included patients had moderate stroke with notable functional impairments affecting daily activities. The mean (SD) age was 62.5 (SD 10.57) years. The experimental group exhibited slightly greater stroke severity than the control group (mean NIHSS score 7.30, SD 1.75 vs 5.75, SD 1.97; P=.01), while other clinical features were comparable between groups.

For the primary outcome, the analysis of FVC revealed significant main effects in both the time (F_1,37=41.12; η²=0.52; P<.001) and time×group interaction effect (F_1,37=4.50; η²=0.12; P=.008), without group effect (F_1,37=4.05; η²=0.10; P=.05; Table 2). The FVC at week 2 increased by 1.12 L (95% CI 0.43‐1.81) in the experimental group and 0.58 L (95% CI 0.02‐1.14) in the control group, a significant difference was observed between the groups at week 2 (LS mean 0.77, 95% CI 0.39‐1.16; η²=0.32; P<.001; Figure 5). The improvements of both groups exceeded the minimal clinically important difference (5%), but the experimental group showed a greater improvement (Table 3).

As for secondary outcome in Tables 2 and 3, there was no significant main effect of group for FEV₁ (F_1,37=2.21; η²=0.06; P=.15), PEF (F_1,37=1.89; η²=0.05; P=.18), whereas MIP (F_1,37=14.49; η²=0.28; P<.001), MEP (F_1,37=10.54; η²=0.20; P=.002), and MBI (F_1,37=7.56; η²=.17; P=.009) reached statistical significance. The main effect of time for FEV₁ (F_0.83,30.71=6.86; η²=0.16; P=.004), PEF (F_0.83,30.71=1.37; η²=0.14; P=.008), MIP (F_1,37=7.66; η²=0.17; P<.001), and MBI (F_1,37=10.07; η²=0.21; P<.001) showed significant differences, but not on MEP (F_0.76,28.12=1.37; η²=0.04; P=.26) at weeks 1 and 2. Significant group×time interactions were found for MIP (F_1,37=7.28; η²=0.16; P=.001), MEP (F_0.76,28.12=4.62; η²=0.11; P=.02), and MBI (F_1,37=3.35; η²=0.08; P=.04), but not for FEV₁ (F_0.83,30.71=1.25; η²=.00; P=0.29) and PEF (F_0.83,30.71=1.66; η²=0.04; P=.20). Compared with baseline, the experimental group exhibited significant increases in FEV₁ (F_1.66,31.54=21.48; η²=0.53; P<.001), PEF (F_1.66,31.54=19.09; η²=0.50; P<.001), MIP (F_2,38=41.48; η²=0.69; P<.001), MEP (F_1.52,28.88=21.59; η²=0.53; P<.001), and MBI (F_2,38=90.42; η²=0.83; P<.001), whereas the control group showed significant increases in FEV₁ (F_1.66,31.54=10.74; η²=0.36; P<.001), MIP (F_2,38=8.50; η²=0.31; P<.001) and MBI (F_2,38=90.42; η²=0.85; P<.001), but not in PEF (F_1.66,31.54=2.99, η²=0.14; P=.07) and MEP (F_1.52,28.88=0.48; η²=0.03; P=.55). Compared with the control group, the experimental group had no baseline differences (P>.05 for FEV₁, PEF, MIP, MEP, and MBI), and showed significant improvements at week 2 in MIP (LS mean 11.38, 95% CI 4.69‐18.08; η²=0.25; P=.001), MEP (LS mean 17.85, 95% CI 9.92‐25.77; η²=0.08; P<.001), and MBI (LS mean 9.92, 95% CI 3.90‐14.53; η²=0.26; P=.001; Figures 6 and 7), as well as in MIP at week 1 (LS mean 10.83, 95% CI 4.12‐17.54; η²=0.23; P=.002). No difference was found between the groups in the FEV₁ and PEF at any assessment time point, which may be due to the short intervention duration and the limited sensitivity of these measures to detect subtle respiratory changes.

This randomized controlled trial demonstrated that a 2-week, hospital-based CMRTP combined with conventional rehabilitation significantly improved respiratory function and daily functional performance compared with conventional rehabilitation alone in patients with moderate acute stroke. Importantly, no serious adverse events were reported, and the mobile-based platform (AIRHUB) was feasible for implementation in inpatient acute stroke care. To our knowledge, this was the first study to establish the efficacy of a mobile-based respiratory program as a safe and effective adjunct to standard care during the acute phase of stroke.

Compared with prior studies [8,43,44], the experimental group demonstrated comparable improvements in respiratory function, respiratory muscle strength, and daily activities in patients with acute stroke. Choi et al [8] reported that a 4-week comprehensive respiratory muscle training program, incorporating air-stacking and respiratory muscle strengthening, led to significant improvements in FVC and MEP in patients with acute stroke. Similarly, Yoo et al [43] found that delivering the same protocol at the bedside over three weeks resulted in increases in FVC, FEV₁, and MBI. In another study, Wei et al [44] found that a 12-week intervention combining respiratory muscle training, pursed-lip breathing, and diaphragmatic breathing improved FVC, FEV₁, respiratory muscle strength, and MBI in patients with stroke.

FVC, as a comprehensive indicator of ventilatory capacity, was particularly sensitive to changes in respiratory muscle performance [43]. Given the prevalence and clinical relevance of respiratory dysfunction in acute stroke, FVC provided an objective and meaningful measure to evaluate the efficacy of respiratory interventions [45]. In this study, both groups showed improvements in FVC after 2 weeks of intervention, but the experimental group achieved greater gains. These intergroup differences were unlikely to be attributed to spontaneous recovery alone but were more plausibly linked to the structured and synergistic design of the CMRTP, encompassing both muscular and neurological mechanisms.

The observed improvements in respiratory function likely resulted from multilevel adaptations involving both central neural remodeling and peripheral muscular strengthening. The acute poststroke phase represented a critical window of heightened neuroplasticity [46], during which axonal sprouting and synaptic reorganization occur at an accelerated rate [47]. Previous studies showed that exercise can facilitate long-term potentiation–like synaptic plasticity, enhancing functions such as memory and fine motor skills [48,49]. Structured and repetitive respiratory exercises—such as diaphragmatic breathing, pursed-lip breathing, and air-stacking—might elicit similar effects by providing continuous sensory and proprioceptive input to cortical and brainstem respiratory centers, reinforcing sensorimotor pathways and supporting circuit-level functional recovery [50]. In addition to synaptic mechanisms, respiratory training might promote neuroplasticity at the molecular level. Exercise was shown to upregulate the expression of mature brain-derived neurotrophic factor, a key modulator of synaptic strength and corticospinal connectivity [51,52]. These effects might enhance the cortical drive to spinal respiratory motor neurons, facilitating compensatory activation and improving voluntary control over respiratory muscles [53-55].

Peripherally, the CMRTP targeted both inspiratory and expiratory muscle groups through focused protocols. Inspiratory exercises and air-stacking primarily strengthened the diaphragm and intercostal muscles [56], whereas pursed-lip and expiratory resistance breathing enhanced the performance of abdominal and accessory expiratory muscles [57-59]. These interventions likely contributed to the significant improvements in MIP and MEP observed in the intervention group. Repeated mechanical loading and neuromuscular activation might have enhanced motor unit recruitment, firing efficiency, and muscle coordination within the respiratory system [60,61]. Together, these central and peripheral adaptations improved respiratory mechanics and endurance, which might have translated into better exercise tolerance and reduced dyspnea. As a result, participants could engage more effectively in physical rehabilitation and daily activities, reflected in the observed improvements in MBI scores.

Despite these positive findings, no significant between-group differences were observed in FEV₁ and PEF. These indices primarily reflect airway patency and expiratory flow dynamics, which are less responsive to short-term interventions and more sensitive to chronic airway function and expiratory control [62]. The 2-week duration of training may have been insufficient to elicit detectable changes in these parameters. Moreover, poststroke impairment in airway coordination—such as reduced glottic control or weakened expiratory reflexes—may further limit responsiveness to early-phase interventions [63,64]. Longer or more targeted expiratory-focused protocols may be needed to achieve improvements in these measures.

Compared with other motor telerehabilitation studies, our findings highlight the value of a simplified, focused intervention suitable for patients with acute stroke. Previous programs targeting upper limb [65], spinal stability [66], or core function [67] have shown functional gains, but often required long durations (8‐12 wk), complex protocols, or real-time supervision, which may limit adherence or scalability. Some participants in these studies reported training fatigue, reduced engagement, or low compliance [66,67]. In contrast, our 2-week respiratory protocol achieved a high adherence rate (96%) with minimal supervision, supported by an app-based system that provided clear instructions and immediate feedback. Rather than using videoconferencing or immersive visual reality [25,66,68], which required constant therapist involvement, our approach relied on routine caregiver support, enhancing feasibility in clinical settings. By focusing on a single, targeted domain—respiratory recovery—our program provided a practical and scalable model for early-phase telerehabilitation in stroke care.

Despite these promising findings, the results must be interpreted in the context of certain limitations. The intervention period was limited to two weeks due to constraints imposed by the Chinese health insurance system, whereas previous studies suggested that respiratory training typically spans at least 4 weeks to maximize benefit [8,59,69]. To address this, the CMRTP was intentionally designed as a high-intensity, twice-daily, multicomponent protocol targeting multiple facets of respiratory function simultaneously. Although the improvements observed in FVC were smaller than those reported in longer trials—for example, a recent meta-analysis reported an average FVC increase of 0.87 L in early stroke [9]—our findings suggest that a condensed, intensive approach could still yield meaningful gains, particularly for patients in the acute phase of stroke.

The AIRHUB platform significantly facilitated the delivery of this program by overcoming temporal and spatial barriers that often impede rehabilitation in the acute phase. It mitigated treatment disruptions caused by limited clinical staff availability or urgent medical needs, ensuring continuous and stable access to respiratory training. Its user-friendly interface, guided multimedia instructions, and integration with nursing oversight enabled patients to complete high-quality training sessions independently and consistently. The hybrid model, combining digital support with in-hospital supervision, likely contributed to the high adherence rate (97.5%) and absence of adverse events. Compared to home-based rehabilitation models, this approach provided greater control over intervention fidelity, which was critical in early-phase recovery.

Nevertheless, several limitations should be acknowledged. First, although outcome assessors were blinded, patient awareness of group allocation might have introduced performance bias, particularly for effort-dependent measures such as MIP and MEP. Second, the relatively short duration of the intervention limited the interpretation of long-term effects. Third, the modest sample size (N=40), while sufficient to detect primary effects, might reduce generalizability and statistical power to detect smaller but clinically meaningful differences.

Recent advances in respiratory rehabilitation included virtual reality biofeedback systems that visualized and quantified respiratory data, offering more effective and engaging training experiences [70,71]. Devices such as Acapella combined with the active cycle of breathing technique improved lung function in perioperative patients with lung cancer [72], and mobile-based intelligent trainers like AeroFit IMT enhanced respiratory muscle strength without inducing fatigue [73]. While these technologies showed promise, their use often required specialized equipment and supervision, which might limit feasibility in acute stroke care [74]. Therefore, incentive spirometry was chosen in this study for its safety, low cost, ease of training, and visual feedback–facilitated adherence [74,75].

Future studies should investigate longer interventions, larger multicenter cohorts, and the integration of advanced or digitally enhanced respiratory devices. Comparative studies of different training intensities, configurations, and delivery models, including home-based telerehabilitation, will be essential to optimize individualized protocols and evaluate scalability in diverse clinical settings.

This randomized controlled trial demonstrated that a 2-week, hospital-based CMRTP combined with conventional rehabilitation significantly improved respiratory function and daily functional performance in patients with moderate acute stroke. The intervention was safe, well-tolerated, and achieved high adherence, suggesting good feasibility for inpatient application. These findings support the integration of mobile-based respiratory training as an effective adjunct to conventional hospital rehabilitation for early respiratory recovery after stroke. Larger multicenter studies with extended follow-up are needed to confirm the long-term benefits and cost-effectiveness of this approach.