Multi parameter respiratory training for pulmonary rehabilitation in AECOPD a randomized controlled trial Authors: Yinan Xing, Guanhua Hou, Wei Li, Zishuo Wang, Limin Zhao Published: 2026 ABSTRACT AbstractEarly pulmonary rehabilitation is recommended for patients with acute exacerbations of chronic obstructive pulmonary disease, yet challenges like monotony and poor adherence remain. This study evaluated a novel multi-parameter respiratory trainer against traditional methods. In a randomized controlled trial, 90 patients were assigned to three groups: multi-parameter device, three-ball trainer, or pursed-lip breathing, alongside standard medication. The primary outcomes were dyspnea, health status, and exercise capacity, with psychological status and lung function as secondary outcomes. Baseline data were comparable. Post-treatment, the multi-parameter device group demonstrated significantly greater improvements in dyspnea, CAT score, six-minute walking distance, and depression and anxiety scores compared to both traditional groups. For lung function, this group also showed superior improvement in FEV1 change versus the pursed-lip breathing group. All three groups showed significant within-group improvements. The intervention was safe, with no adverse events reported. Under supervised inpatient conditions, the multi-parameter respiratory training device is safe and effective for short-term pulmonary rehabilitation in AECOPD patients, offering superior improvements in patient-centered outcomes compared with conventional methods. However, the evidence is limited to supervised inpatient settings, and the device should be considered a complementary option to, not a replacement for, basic respiratory training when resources permit.Keywords: AECOPD, COPD, Pulmonary rehabilitation, Respiratory therapy, Respiratory training deviceSubject terms: Diseases, Health care, Medical research Introduction Chronic Obstructive Pulmonary Disease (COPD) is a common and preventable chronic airway disease characterized by persistent airflow limitation and corresponding respiratory symptom1. It features a long disease course, frequent acute exacerbations, and various comorbidities. Patients often experience progressively worsening dyspnea, decreased exercise tolerance, and deteriorated quality of life, which not only severely impairs individual health but also imposes a heavy economic burden on families and society2. Chronic Obstructive Pulmonary Disease (COPD) ranks as the fourth leading cause of death globally, accounting for 3.5 million deaths in 2021, which constitutes approximately 5% of the total global mortality3. Studies have projected that by 2050, there will be 15.6 billion episodes of acute exacerbation of COPD worldwide, with the cumulative direct costs estimated to reach 24.35 trillion US dollars and indirect costs totaling 15.43 trillion US dollars4. COPD treatment includes pharmacological and non-pharmacological interventions. Pulmonary Rehabilitation (PR) falls under non-pharmacological intervention. The concept of PR was first proposed in 19725. The GOLD 2025 guidelines further reinforce the core role of PR as a non-pharmacological treatment for COPD, explicitly listing it as a Class IA recommended intervention6. PR, a multi-component intervention encompassing exercise training, breathing exercises, and health education, has been demonstrated to significantly improve patients’ exercise capacity, alleviate dyspnea, and enhance quality of life7. An analysis incorporating 15 studies involving 1047 subjects confirmed that PR exercise significantly improved FEV1, FEV1/FVC, FEV1%, increased the 6-minute walk test (6MWT) distance, and significantly enhanced lung function and exercise capacity in COPD patients8. Respiratory training is a key component of PR, including inspiratory muscle training, expiratory muscle training, and diaphragmatic training9. Respiratory trainers are assistive devices for this purpose. The three-ball breathing trainer is currently widely used. It helps patients gradually increase inspiratory or expiratory muscle strength by providing adjustable resistance, offering advantages like intuitive feedback and simple operation10. However, its structure and function are relatively limited; it cannot tailor training based on individual patient differences, and the training process can be monotonous, leading to low patient participation and adherence. One study showed that only about 31.2% of Acute Exacerbation of Chronic Obstructive Pulmonary Disease (AECOPD) patients participated in PR, with merely 15% completing the full course11. In a study by Milner, 77.42% of the 31 enrolled AECOPD patients refused PR12. Based on this background, this study utilized a multi-parameter respiratory trainer. It is a personalized respiratory training device integrating multiple functions such as respiratory function assessment, general training mode, positive pressure training mode, voice guidance, and a combination of various breathing training methods, while displaying multiple parameters. Its core innovation is the ability to assess the user’s COPD grade via an initial expiratory test or questionnaire and subsequently set personalized inspiratory pressure and expiratory resistance. It also guides and controls the patient’s breathing frequency through voice rhythm, thus achieving personalized PR. This randomized controlled trial systematically evaluated the efficacy and safety of this multi-parameter respiratory training device compared to a traditional three-ball breathing trainer and conventional pursed-lip breathing with diaphragmatic breathing in PR for AECOPD patients, to explore its potential clinical application value. Materials and methods From October 2024 to May 2025, 90 patients diagnosed with AECOPD were recruited from the Department of Respiratory and Critical Care Medicine at Henan Provincial People’s Hospital. Eligible patients were randomly assigned to one of three experimental groups: Group A (multi-parameter respiratory training device), Group B (three-ball breathing trainer), or Group C (pursed-lip breathing with diaphragmatic breathing). Inclusion criteria: (1) Males and females aged 18–75 years; (2) Patients meeting COPD diagnostic criteria per the Global Initiative for Chronic Obstructive Lung Disease, with airflow limitation severity at GOLD grades 2–3, assessed during an acute exacerbation; (3) No history of PR. Exclusion criteria: (1) Lack of informed consent; (2) Comorbidities like bronchiectasis, diabetes, or severe cardiac, hepatic, renal systemic diseases; (3) Lower limb dysfunction preventing 6MWT completion; (4) Giant pulmonary bullae; (5) Malignant hypertension; (6) Cognitive impairment; (7) Concurrent malignant tumors; (8) History of acute cerebral infarction, cerebral hemorrhage, or other severe cardiovascular/cerebrovascular events within the past 3 months. Eligible patients were randomly divided into Groups A, B, and C (n = 30 each). All groups received conventional pharmacological treatment. Rehabilitation training lasted 7 days, twice daily for 15 min per session, supervised by trained staff. Sessions were stopped if discomfort occurred. Group A used the multi-parameter device described above. Training followed an intermittent pattern: 10 breaths per set, 30–60 s rest between sets, repeated to complete 15 min. Group B used a standard three-ball trainer. Patients inhaled deeply to raise the balls. Training consisted of 2–3 sets of 10–15 breaths per session, totaling 15 min. Intensity was self-paced; no fixed resistance was applied. Group C performed pursed-lip breathing combined with diaphragmatic breathing (inspiration: expiration = 1:2–1:3, respiratory rate 8–10 breaths/min) continuously for 15 min per session. While therapeutic intensity could not be standardized across modalities, the dose was the same for all groups, ensuring that comparisons reflect modality-specific efficacy rather than differences in treatment exposure. This study was reviewed and approved by the Medical Ethics Committee of Henan Provincial People’s Hospital (Approval No. (2023) Ethics Review No. (164)). All procedures were conducted in accordance with the ethical standards of the institutional research committee and with the 1964 Helsinki Declaration and its later amendments. Written informed consent was obtained from all individual participants prior to their inclusion in the study. This trial was registered at the Chinese Clinical Trial Registry (ChiCTR), a primary registry of the WHO International Clinical Trials Registry Platform. The registration identifier is ChiCTR2600116557, and the publicly accessible URL is: http://www.chictr.org.cn/bin/project/edit?pid=299539. The date of registration was 2026-01-12. This randomized controlled trial was designed, conducted, and reported in accordance with the CONSORT 2025 Statement. The completed CONSORT 2025 checklist is provided as supplementary material. Randomization: A computer-generated random number sequence was utilized to allocate participants in a 1:1:1 ratio to Groups A, B, and C using block randomization. (block size set to 6). The allocation scheme was sealed in opaque, sequentially numbered envelopes and kept by an independent researcher not involved in participant recruitment or intervention delivery. Envelopes were opened in sequential. order after participants completed baseline assessments. Blinding: This study employed assessor blinding. Researchers responsible for measuring outcome indicators (lung function, 6MWD) and data analysis remained blinded to group allocation. Due to the nature of the interventions, blinding of participants and intervention providers was not possible. Sample Size Estimation: The 6MWD was used as the primary indicator for calculation. Based on preliminary pilot studies and relevant literature, it was assumed that Group A (multi-parameter device) would achieve a mean improvement in 6MWD 50 m greater than Group C (conventional training), with a pooled standard deviation of 45 m. Setting a significance level α = 0.05 (two-sided) and a power (1-β) of 0.90, calculation using PASS 2021 software indicated a requirement of 23 patients per group. Accounting for an estimated dropout rate of 25%, the final sample size was set at 30 patients per group, totaling 90 patients. Equipment Multi-parameter respiratory training device The experimental device was a self-developed multi-parameter respiratory trainer (Patent No.: ZL202010527409.4) from Henan Provincial People’s Hospital. Training Procedure: Before training, the patient assumes a comfortable sitting position, places a disposable breathing circuit mouthpiece in the mouth, and uses a nose clip to occlude the nasal passages, ensuring oral breathing for accurate collection of respiratory signals. After starting the device and entering the patient’s basic information (age, sex, height, weight), the system guides the patient through an initial forced expiratory assessment. Following voice prompts, the patient first inhales deeply to total lung capacity and then exhales forcefully and rapidly to residual volume with maximal effort. The system automatically measures and records the following pulmonary function parameters: FVC, FEV1, FEV1/FVC, peak expiratory flow (PEF), and forced expiratory flow at 25%, 50%, and 75% of FVC (FEF25, FEF50, FEF75). This assessment is repeated 2–3 times, and the system automatically selects the best value as the assessment result. Based on the assessment results and combined with manually entered GOLD grade (2 or 3), the system automatically calculates and generates individualized initial training loads: Inspiratory pressure support level: Set based on FEV1 as a percentage of predicted value. When FEV1% predicted < 50% (GOLD grade 3), the initial inspiratory pressure support level is set at 8–12 cmH₂O; when FEV1% predicted ≥ 50% (GOLD grade 2), the initial inspiratory pressure support level is set at 5–8 cmH₂O. Expiratory resistance value: Set based on measured values of PEF and FEF50. Expiratory resistance is divided into five levels (Level 1 being the lowest resistance, Level 5 the highest), with an initial setting at Level 3 (moderate resistance). These parameter settings remained fixed throughout the 7-day intervention period to ensure consistency of the training protocol. The researcher recorded the initial parameter settings for each patient and verified the device parameters before each daily training session to ensure accuracy. Subsequent adjustments could be made manually based on patient tolerance: if the patient showed no significant fatigue and tolerated the training well, the resistance level could be increased by one level; if the patient experienced significant discomfort, dyspnea exacerbation, or oxygen desaturation, the resistance level could be decreased by one level. The training mode is selected based on the patient’s inspiratory muscle strength assessment: General training mode: This mode is suitable for patients with adequate inspiratory muscle strength who can overcome inspiratory resistance independently to complete deep inhalation. The criteria for selection are: during the initial forced expiratory assessment, the patient can independently complete deep inhalation without significant involvement of accessory respiratory muscles, or the researcher assesses the inspiratory effort as ≥ moderate. In this mode, the device provides no positive-pressure assistance during inspiration; only the preset inspiratory resistance value is applied for resistive inspiratory training. During expiration, the preset expiratory resistance value is applied for resistive expiratory training. Positive-pressure training mode: This mode is suitable for patients with obvious inspiratory muscle weakness. The criteria for selection are: during the initial forced expiratory assessment, the patient shows significant reliance on accessory respiratory muscles during deep inhalation, demonstrates insufficient inspiratory effort, or has a history of respiratory muscle fatigue (e.g., shortness of breath during daily activities, reliance on pursed-lip breathing). In this mode, the device delivers continuous, controllable positive-pressure airflow during inspiration based on the preset inspiratory pressure value, assisting the patient in completing the inspiratory maneuver and reducing inspiratory muscle load. During expiration, expiratory resistance training is retained. The device’s core mechanism involves controlled positive pressure inspiration reducing premature small airway closure, promoting residual gas expulsion, decreasing dynamic hyperinflation, improving V/Q ratio, coordinating respiratory muscles, and relieving muscle fatigue13–15. Device humidification reduces airway irritation, alleviating cough and chest tightness. Furthermore, the COPD grading function allows precise pre-training assessment16, enabling personalized resistance and pressure settings, key to its enhanced symptom improvement. Three-ball breathing trainer The three-ball breathing trainer is the most widely used basic respiratory training device in clinical practice for COPD. The device mainly consists of a sterile mouthpiece, a transparent cylindrical chamber, three-level suspended balls, airflow scale markings, and an anti-slip base, adopting an integrated design. The mouthpiece can be replaced individually, and the main body of the device can be reused after standard disinfection, making it suitable for both clinical and home-based rehabilitation settings. Its core working principle is that the airflow generated by the patient‘s spontaneous inhalation or exhalation causes the balls inside the chamber to suspend. The number and height of the suspended balls intuitively reflect the patient’s respiratory airflow volume and pressure. By increasing the workload on the respiratory muscles, it enhances respiratory muscle strength and endurance, thereby improving pulmonary ventilation function. Observation indicators Dyspnea: Dyspnea severity was assessed using the modified Medical Research Council (mMRC) dyspnea scale. This is a 5-point (0–4) self-rated questionnaire based on the level of physical activity that provokes breathlessness. Higher scores indicate more severe dyspnea. The minimal clinically important difference (MCID) for mMRC is generally accepted as a reduction of 1 point17,18. Health-Related Quality of Life: Health status was evaluated using the COPD Assessment Test (CAT). This is an 8-item patient-completed questionnaire covering symptoms (cough, sputum, chest tightness, dyspnea) and activity limitations (home activities, confidence, sleep, energy). Each item is scored from 0 to 5, yielding a total score ranging from 0 to 40. Higher scores represent a greater impact of COPD on daily life and worse health status. The MCID for CAT has been established as a 2-point improvement18–20. Exercise Capacity: Functional exercise capacity was measured using the 6MWD test. The test was conducted twice in a 30-meter flat, enclosed corridor following the American Thoracic Society (ATS) guidelines, including the use of standardized encouragement. The best of the two attempts was recorded for analysis. The MCID for 6MWD is widely accepted as 26–35 m, with 30 m commonly used as the clinical threshold21. Psychological Status: Anxiety and depression symptoms were assessed using the Self-Rating Anxiety Scale (SAS) and Self-Rating Depression Scale (SDS), respectively. Both are 20-item self-administered questionnaires. Each item is scored on a 4-point Likert scale based on the frequency of symptoms. The raw total score (ranging 20–80) is converted to a standardized index score (ranging 25–100). For both scales, higher scores indicate more severe anxiety or depressive symptoms. Lung Function: Pulmonary function was assessed via spirometry, measuring forced vital capacity (FVC), forced expiratory volume in 1 s (FEV1), and the FEV1/FVC ratio. Measurements were expressed in absolute values (Liters) and as a percentage (%) for the ratio. Safety: Safety was evaluated by monitoring the incidence of adverse events throughout the intervention period. Adverse events were defined as any untoward medical occurrence, including but not limited to worsened dyspnea, chest pain, palpitations, fatigue, and dizziness. Statistical analysis Statistical analysis was performed using SPSS 27.0. Continuous data were presented as mean ± standard deviation(x̄±s), and categorical data as frequencies. Intergroup comparisons were analyzed using one-way ANOVA with LSD or Dunnett’s T3 post-hoc tests. Intragroup comparisons were performed using paired t-tests. Mean differences and 95% confidence intervals were calculated from the ANOVA model. Categorical variables were compared using the chi-square test, with P < 0.05 as statistically significant. This analysis was conducted based on the intention-to-treat (ITT) principle, which mandates that all randomized participants be included in the analysis according to their original group allocation22. In this study, all 90 randomized subjects completed the full 7-day intervention and the endpoint assessments on day 8, with no dropouts, withdrawals, or protocol violations. Consequently, there were no missing data for any of the primary or secondary outcome measures. Given the complete adherence and follow-up, the ITT population is identical to the per-protocol (PP) population in this specific trial. Therefore, the reported results simultaneously satisfy both ITT and PP criteria, and the analysis is aligned with the CONSORT 2025 recommendations for minimizing bias. Results Patient flow and baseline characteristics A total of 90 patients were screened, all of whom met the eligibility criteria and were subsequently randomized into three groups: Group A (multi-parameter device, n = 30), Group B (three-ball trainer, n = 30), and Group C (pursed-lip breathing, n = 30). All randomized patients completed the 7-day intervention protocol and follow-up assessments, with no dropouts or exclusions from the analysis (Fig. 1). Flow chart of the participants’ progress during the study. 90 AECOPD patients were randomized into Groups A, B, C (n = 30). No significant differences were found in age, BMI, gender, smoking history, comorbidities, or exacerbations in the past year (P > 0.05), indicating comparable baseline data (Table 1). Table 1Comparison of baseline characteristics among the three groups.VariableABCP-valueAge(year)64.63 ± 9.2964.20 ± 8.6066.67 ± 5.740.449BMI (kg/m²)24.17 ± 3.3424.68 ± 3.9524.02 ± 4.390.795Gender (n%)men27 (90.0)26 (86.7)28 (93.3)0.69women3 (10.0)4 (13.3)2 (6.7)Smoke (n%)yes25 (83.3)24 (80.0)24 (80.0)0.93no5 (16.7)6 (20.0)6 (20.0)Number of complication(n%)013 (43.3)14 (46.7)12 (40.0)0.97818 (26.7)10 (33.3)10 (33.3)27 (23.3)5 (16.7)6 (20.0)≥ 32 (6.7)1 (3.3)2 (6.7)Number of acute exacerbations in the past year (n%)07 (23.3)6 (20.0)5 (16.7)0.99115 (50.0)17 (56.7)16 (53.3)26 (20.0)5 (16.7)6 (20.0)≥ 32 (6.7)2 (6.7)3 (10.0)Data are presented as mean ± standard deviation(x̄±s), n (%). Continuous variables were analyzed by one-way ANOVA, categorical variables by chi-square test. The significance level was set at P < 0.05. Comparison of baseline characteristics among the three groups. Data are presented as mean ± standard deviation(x̄±s), n (%). Continuous variables were analyzed by one-way ANOVA, categorical variables by chi-square test. The significance level was set at P < 0.05. Pre-treatment outcome measure Consistent with the baseline demographics, the pre-treatment assessments of all outcome measures, including mMRC, CAT, 6MWD, SDS, SAS, FVC, FEV1, FEV1/FVC showed no significant differences among the three groups (P > 0.05). This further confirmed the homogeneity of the groups prior to the intervention (Table 2). Table 2Comparison of pre-treatment outcome measures among the three groups.VariableABCP-valuemMRC3.00(2.00,3.00)3.00(2.00,3.00)3.00(2.00,3.00)0.981CAT22.93 ± 6.4323.30 ± 5.3523.00 ± 6.200.9696MWD(m)326.47 ± 65.20324.70 ± 47.35325.77 ± 45.720.992SDS33.03 ± 6.3332.50 ± 5.1932.73 ± 5.320.935SAS33.12 ± 6.1431.73 ± 5.3832.57 ± 3.970.591FVC(L)3.03 ± 0.902.91 ± 0.693.00 ± 0.540.817FEV1(L)1.46 ± 0.591.40 ± 0.371.45 ± 0.520.903FEV1/FVC(%)48.43 ± 12.4549.19 ± 11.0848.81 ± 14.370.974mMRC, modified Medical Research Council dyspnea scale; CAT, COPD Assessment Test; 6MWD, 6-minute walking distance; SDS, Self-Rating Depression Scale; SAS, Self-Rating Anxiety Scale; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 second; FEV1/FVC, ratioof forced expiratory volume in 1 second to forced vital capacity.Data are presented as mean±standard deviation (x̄±s) or median (IQR) for mMRC. P-values were derived from one-way ANOVA or Kruskal-Wallis test. The significance level was set at P<0.05. Comparison of pre-treatment outcome measures among the three groups. mMRC, modified Medical Research Council dyspnea scale; CAT, COPD Assessment Test; 6MWD, 6-minute walking distance; SDS, Self-Rating Depression Scale; SAS, Self-Rating Anxiety Scale; FVC, forced vital capacity; FEV1, forced expiratory volume in 1 second; FEV1/FVC, ratioof forced expiratory volume in 1 second to forced vital capacity. Data are presented as mean±standard deviation (x̄±s) or median (IQR) for mMRC. P-values were derived from one-way ANOVA or Kruskal-Wallis test. The significance level was set at P<0.05. Effects of intervention: within-group comparisons Following the 7-day rehabilitation program, significant improvements were observed within all three groups across nearly all outcome measures. These within-group improvements were highly statistically significant (P < 0.01) for CAT, 6MWD, SDS, SAS, FEV1, and FEV1/FVC in all groups. Significant improvements in FVC were also noted in Group A (P = 0.040) and Group B (P = 0.019), but not in Group C (P = 0.170). The mMRC scores decreased significantly in all groups (P < 0.01). (Table 3). Table 3Intra-group comparisons of outcome measures before and after treatment.VariableGroupBaselinePostP-valuemMRCA3.00(2.00,3.00)1.00(1.00,2.00)< 0.001B3.00(2.00,3.00)2.00(2.00,2.00)< 0.001C3.00(2.00,3.00)2.00(2.00,3.00)0.002CATA22.93 ± 6.4313.53 ± 4.15< 0.001B23.30 ± 5.3516.93 ± 4.79< 0.001C23.00 ± 6.2019.40 ± 5.48< 0.0016MWD(m)A326.47 ± 65.20397.50 ± 49.52< 0.001B324.70 ± 47.35364.33 ± 49.43< 0.001C325.77 ± 45.72347.67 ± 41.22< 0.001SDSA33.03 ± 6.3326.43 ± 5.22< 0.001B32.50 ± 5.1928.53 ± 5.18< 0.001C32.73 ± 5.3231.23 ± 5.39< 0.001SASA33.12 ± 6.1427.13 ± 4.20< 0.001B31.73 ± 5.3828.53 ± 4.58< 0.001C32.57 ± 3.9731.43 ± 4.09< 0.001FVC(L)A3.03 ± 0.903.22 ± 0.870.040B2.91 ± 0.693.12 ± 0.810.019C3.00 ± 0.543.16 ± 0.910.170FEV1(L)A1.46 ± 0.591.76 ± 0.61< 0.001B1.40 ± 0.371.65 ± 0.38< 0.001C1.45 ± 0.521.60 ± 0.48< 0.001FEV1/FVC(%)A48.43 ± 12.4554.46 ± 11.19< 0.001B49.19 ± 11.0854.32 ± 10.87< 0.001C48.81 ± 14.3751.83 ± 13.480.002Data are presented as mean±standard deviation (x̄±s) or median (IQR) for mMRC. Within-group comparisons were performed using paired t-test, Wilcoxon signed-rank test for mMRC. The significance level was set at P < 0.05 Intra-group comparisons of outcome measures before and after treatment. Data are presented as mean±standard deviation (x̄±s) or median (IQR) for mMRC. Within-group comparisons were performed using paired t-test, Wilcoxon signed-rank test for mMRC. The significance level was set at P < 0.05 Between-group comparisons of efficacy The mean differences and 95% confidence intervals for between-group comparisons are presented in Table 4. The comparison of the absolute improvement (Δ values) between groups revealed a clear and consistent gradient of efficacy. The multi-parameter respiratory training device (Group A) demonstrated significantly greater improvements in all patient-centered clinical outcomes compared to both the three-ball trainer group (Group B) and the pursed-lip breathing group (Group C). This superior efficacy was evident in ΔCAT, Δ6MWD, ΔSDS, ΔSAS, and ΔmMRC (all P < 0.01 for A vs. C, and all P ≤ 0.004 for A vs. B). Notably, the improvement in 6MWD in Group A and Group B exceeded the established MCID of 30 m, while the improvement in Group C did not reach this clinically meaningful threshold. Table 4Between-group comparisons of changes in outcome measures (∆ values).VariableABCMean difference (95% CI)P-value (A vs. B)Mean difference (95% CI)P-value (A vs. C)Mean difference (95% CI)P-value (B vs. C)∆mMRC− 2.00 (− 2.00, − 1.00)− 1.00 (− 1.00, 0.00)− 1.00 (− 1.00, 0.00)0.004< 0.0010.085∆CAT9.40 ± 3.786.37 ± 2.703.60 ± 1.553.03 (1.35–4.71)< 0.0015.80 (4.12–7.48)< 0.0012.77 (1.49–4.05)< 0.001∆6MWD(m)71.03 ± 41.5239.63 ± 15.5421.90 ± 8.8031.40 (14.57–48.23)< 0.00149.13 (32.30–65.96)< 0.00117.73 (10.54–24.92)< 0.001∆SDS6.60 ± 4.353.97 ± 2.761.50 ± 1.082.63 (0.78–4.48)0.0015.10 (3.25–6.95)< 0.0012.47 (1.35–3.59)< 0.001∆SAS5.98 ± 4.543.20 ± 2.271.13 ± 1.142.78 (0.94–4.62)0.0024.85 (3.01–6.69)< 0.0012.07 (1.07–3.07)< 0.001∆FVC(L)0.20 ± 0.490.21 ± 0.460.16 ± 0.63− 0.01 (− 0.25–0.23)0.9320.04 (− 0.20–0.28)0.7830.05 (− 0.23–0.33)0.724∆FEV1(L)0.30 ± 0.230.25 ± 0.170.15 ± 0.200.05 (− 0.05–0.15)0.3180.15 (0.05–0.25)0.0050.10 (0.00–0.20)0.051∆FEV1/FVC(%)6.03 ± 2.155.13 ± 2.453.02 ± 4.910.90 (− 0.23–2.03)0.1183.01 (1.45–4.57)< 0.0012.11 (0.37–3.85)0.018Data are presented as mean±standard deviation (x̄±s) or median (IQR) for ΔmMRC. Between-group comparisons are presented as mean difference (95% confidence interval) followed by P-value. Δvalues represent the absolute magnitude of improvement from baseline to post-treatment. Negative Δ values for mMRC indicate a greater reduction in dyspnea. Inter-group comparisons of continuous Δ values were analyzed by one-way ANOVA with LSD post-hoc test, except for ΔmMRC which was analyzed by the Kruskal-Wallis test followed by Mann-Whitney U test with Bonferroni correction. The significance level was set at P < 0.05. Between-group comparisons of changes in outcome measures (∆ values). Data are presented as mean±standard deviation (x̄±s) or median (IQR) for ΔmMRC. Between-group comparisons are presented as mean difference (95% confidence interval) followed by P-value. Δvalues represent the absolute magnitude of improvement from baseline to post-treatment. Negative Δ values for mMRC indicate a greater reduction in dyspnea. Inter-group comparisons of continuous Δ values were analyzed by one-way ANOVA with LSD post-hoc test, except for ΔmMRC which was analyzed by the Kruskal-Wallis test followed by Mann-Whitney U test with Bonferroni correction. The significance level was set at P < 0.05. For lung function parameters, the improvement in ΔFEV1 was significantly greater in Group A than in Group C (P = 0.005), and the improvements in ΔFEV1/FVC in both Group A and Group B were superior to Group C (P < 0.001 and P = 0.018, respectively). No significant differences were found between the two device-based groups in ΔFEV1 and ΔFEV1/FVC. There were no significant inter-group differences in the change of FVC (P > 0.05). (Table 4). Safety No significant adverse events occurred in any group, which is consistent with most research findings23,24. confirming the good safety of early PR in hospitalized AECOPD patients. Discussion In this study, the multi-parameter device group showed greater reductions in mMRC and CAT scores than the three-ball trainer and pursed-lip breathing groups, suggesting superior dyspnea relief and symptom reduction. This finding is consistent with previous studies. Costi17 et al. and Filizola18 et al. both demonstrated that device-guided breathing training alleviates dyspnea by reducing dynamic hyperinflation and improving the ventilation/perfusion ratio. The significant CAT score reduction relates to positive feedback during training. Study indicated that self-efficacy in COPD patients aids dyspnea relief25. Unlike traditional training, the device’s real-time display of 14 parameters, dynamic curves, and visualizations allow patients to intuitively perceive effects, enhancing confidence and self-efficacy, improving acceptance, sleep quality, and daily activity capacity. The superior efficacy of Group A can be primarily attributed to the following integrated features: controllable positive-pressure inspiration reduces inspiratory muscle load, adjustable expiratory resistance strengthens expiratory muscle power, voice guidance standardizes breathing rhythm, and real-time visual feedback consolidates training effects by enhancing self-efficacy. In contrast, the three-ball trainer provides only inspiratory resistance training, and pursed-lip breathing relies entirely on the patient’s subjective effort; both lack the synergistic effect of such an integrated intervention. 6MWD is a core exercise tolerance indicator. The multi-parameter device group’s 6MWD improvement was superior, consistent with findings26 that PR improves tolerance via enhanced muscle function. AECOPD patients often have reduced tolerance due to respiratory muscle fatigue and peripheral muscle dysfunction27. The device enhances expiratory muscle strength through adjustable expiratory resistance, while reducing inspiratory muscle load and improving inspiratory muscle function via positive-pressure inspiration, thereby achieving synergistic training of both inspiratory and expiratory muscles. In contrast, the three-ball trainer targets only the inspiratory muscles. Given that AECOPD patients often present with dysfunction of both inspiratory and expiratory muscles, this dual-phase training mode, which involves positive-pressure assistance during inspiration and resistive loading during expiration, may more comprehensively address their pathophysiological characteristics. In addition, voice guidance helps standardize breathing patterns and reduce exercise-induced dyspnea. Together, these factors explain why Group A outperformed Groups B and C. Anxiety/depression prevalence is higher in AECOPD patients, negatively correlating with adherence28. The multi-parameter device group showed the greatest SDS/SAS score reductions, indicating psychological state improvement. Personalized settings reduce discomfort, alleviating fear of symptom worsening. Dynamic feedback and visualization allow clear progress perception. This finding is consistent with the study by Farver-Vestergaard29 et al., which demonstrated that mindfulness-based training combined with pulmonary rehabilitation significantly improved anxiety and depression symptoms in COPD patients. The real-time visual feedback provided by our device may play a role similar to mindfulness training: by allowing patients to clearly perceive their progress with each breath, it enhances their confidence in disease control and self-efficacy, thereby alleviating negative emotions. Previous research has also confirmed that self-efficacy is an important protective factor for psychological status in COPD patients28. The advantage of Group A in this regard may be attributed to its personalized settings, which reduce initial discomfort, and to real-time feedback, which provides quantifiable evidence of progress, in stark contrast to the lack of objective feedback in Groups B and C. Although FEV1 and FEV1/FVC improved significantly within all groups, inter-group analysis showed the multi-parameter group was only superior to the pursed-lip group in ΔFEV1 and ΔFEV1/FVC, not the three-ball group. This likely stems from device-based training characteristics. Both devices provide objective resistance loads and clear targets with immediate visual feedback, enhancing patient motivation, unlike pursed-lip breathing relying on subjective effort. This finding is consistent with the study by Wang30, which showed that device-assisted breathing training was superior to simple breathing exercises in improving airway function, but the differences between different devices were not evident over a short-term intervention. No significant inter-group differences in ΔFVC were observed, consistent with previous studies31. FVC reflects lung expansile capacity, dependent on lung elastic recoil recovery and thoracic mobility. During exacerbations, reversible factors like secretions and edema are controlled medically, but lost elastic recoil in COPD is irreversible, limiting short-term volume improvement from breathing exercises. During an acute exacerbation, pulmonary function typically improves as part of the natural recovery process in response to standard pharmacological therapy31. The within-group improvements in FEV1 and FVC across all three groups likely reflect, at least in part, this natural resolution of acute inflammation and bronchoconstriction. While the between-group differences suggest an additional benefit of device-based training, attributing the entirety of the lung function improvement to the breathing interventions would be inappropriate. The short intervention period may also affect lung function changes, suggesting that longer training might be needed to detect differences between devices. The fact that Group A was superior to Group C, but not to Group B, only in FEV1 suggests that providing objective resistive load and visual feedback may be the key drivers of short-term airway function improvement, whereas the additional positive-pressure assistance and voice guidance of the multi-parameter device did not translate into extra benefits in pulmonary function indices within the 7-day intervention period. Such benefits might require a longer training duration to emerge. It is noteworthy that while all three groups demonstrated statistically significant improvements in the primary outcome measures, the assessment of clinical significance requires the application of the MCID. The MCID is the smallest change patients perceive as beneficial21. In terms of dyspnea improvement, Group A achieved a median reduction of 2 points, whereas Groups B and C each achieved a 1-point reduction. This suggests that only the improvement in Group A (-2 points) substantially exceeded the MCID threshold. Regarding exercise capacity, Group A demonstrated a mean improvement of 71.03 m; Group B improved by 39.63 m, meeting the MCID; and Group C improved by only 21.90 m, which did not reach the threshold for clinically meaningful improvement. In health status, all three groups surpassed this threshold, indicating that regardless of the breathing training method used, patients achieved clinically meaningful improvements in quality of life. However, the magnitude of benefit was substantially greater in Group A. For the SDS/SAS, no universally accepted MCID thresholds are currently available. Therefore, although Group A showed significantly greater improvements in anxiety and depression compared to the control groups, the clinical significance of these differences requires further validation. A notable demographic characteristic of this study is the marked male predominance. This gender imbalance, while consistent with the epidemiology of COPD in China where historically higher smoking rates and occupational exposures among men have led to greater disease prevalence32, nonetheless represents a significant limitation. Previous research has suggested that women with COPD may experience different symptom profiles, greater dyspnea burden, and potentially differential responses to PR compared to men33. Consequently, the generalizability of our findings to female AECOPD patients is uncertain. Future studies should specifically recruit adequate numbers of female participants to enable gender-stratified analyses and ensure that the benefits of multi-parameter respiratory training are applicable across both sexes. While the improvements observed in this study are encouraging, their magnitude warrants critical examination of potential contributing factors beyond the intervention itself. First, the natural clinical recovery during inpatient treatment for AECOPD likely contributed to the improvements observed across all groups. Standard pharmacological therapy typically leads to resolution of acute inflammation, improved airflow, and symptom relief over the course of hospitalization31. Second, repeated 6MWD testing may be subject to learning effects. Patients unfamiliar with the test often perform better on subsequent attempts due to improved familiarity and pacing strategies34. Third, the absence of participant blinding introduces potential performance and expectation biases35. This is particularly relevant for subjective outcomes (CAT, mMRC, SDS, SAS), which are highly susceptible to patient expectations and motivation. Patients in the device group were aware they were using a novel technology with advanced feedback features, which may have heightened their expectations and led to greater effort during training and testing, potentially influencing these subjective assessments. Fourth, regression to the mean may partially explain the improvements observed36. Patients were enrolled during an acute exacerbation, a period of maximal symptom severity. Some improvement toward their baseline stable state would be expected regardless of intervention. However, the between-group differences mitigate this concern, as regression to the mean would affect all groups similarly. The intervention in this study was delivered in a highly supervised inpatient setting, with training sessions guided throughout by uniformly trained research staff and patient adherence reaching 100%. While such idealised conditions ensured internal validity, they differ considerably from the real-world challenges faced in pulmonary rehabilitation. The uptake of pulmonary rehabilitation after hospital discharge for an acute exacerbation of COPD is generally low. A study reported that less than 3% of patients received pulmonary rehabilitation within one year of discharge, and less than 2% within six months37. Data similarly indicate that fewer than 10% of patients complete outpatient pulmonary rehabilitation following an AECOPD hospitalisation38. In the home-based setting, adherence also faces serious difficulties, influenced mainly by the severity of dyspnoea, the level of social support, and self-efficacy39. The multi-parameter respiratory trainer has tangible clinical advantages. Automated parameter setting reduces the training burden on healthcare staff, and the voice guidance together with real-time feedback may improve patient adherence, thereby addressing a key barrier in current pulmonary rehabilitation practice. However, these advantages have so far been demonstrated only in a highly supervised inpatient environment; their applicability outside this setting remains to be confirmed. The supervised training protocol used in this study cannot be directly transferred to outpatient or home-based rehabilitation. Moreover, the relatively high acquisition cost of the device may limit its dissemination in resource-poor areas, and issues such as maintenance requirements, technical support, and basic staff training are practical considerations that need careful evaluation. It is worth emphasising that, although this study confirmed the superiority of the multi-parameter device across several outcome measures, all three groups achieved clinically meaningful improvements. All three groups exceeded the minimal clinically important difference for the CAT score, and the improvement in 6-minute walking distance reached or surpassed the 30-metre threshold in Groups A and B. This means that even simple breathing interventions, such as the three-ball trainer or pursed-lip breathing, can bring perceptible benefits to patients during an AECOPD hospitalisation. Consequently, limited access to the multi-parameter device should not be equated with a lack of effective pulmonary rehabilitation options. On the contrary, basic breathing exercises should be promoted as standard care across different levels of healthcare facilities; when resources allow, advanced devices can serve as a complementary or step-up choice to provide additional benefits for patients. Based on the above analysis, the superior efficacy observed in this study should be understood as evidence of short-term (7-day) benefit under supervised inpatient conditions. There is currently insufficient evidence to support extrapolating these results directly to unsupervised settings such as outpatient clinics, community programmes, or home environments, or to assume long-term benefits from them. This study does not demonstrate the feasibility, effectiveness, or cost-effectiveness of the device in unsupervised settings. Future research should adopt multicentre designs, extend follow-up periods, conduct cost-effectiveness analyses, develop simplified protocols suitable for home-based rehabilitation, and explore telehealth-supported home care models in order to fully evaluate the value of this device in real-world practice. This study supports the efficacy of the multi-parameter device but has limitations: single-center design with 90 patients potentially introduces selection bias; lack of subgroup analysis based on key clinical features (e.g., GOLD 2 vs. 3 differences); short training duration limited by hospitalization, preventing assessment of long-term effects. Future studies should expand sample size, conduct subgroup analyses, and explore long-term outcomes via remote supervision. What’s more, this study has a methodological limitation. The statistical analysis involved multiple outcome measures and multiple between-group comparisons without adjustment for multiplicity. The use of LSD post-hoc tests, while sensitive for detecting clinically meaningful differences, increases the risk of Type I error inflation. However, the consistent pattern of findings across multiple related outcomes—dyspnea, exercise capacity, quality of life, and psychological well-being—and the fact that improvements exceeded established MCID thresholds provide reassurance that the observed effects are not solely attributable to chance. Future confirmatory studies should prespecify a limited number of primary outcomes and incorporate appropriate multiplicity adjustments. Conclusions This randomized controlled trial indicates that, under supervised inpatient conditions, the multi-parameter respiratory training device is safe and effective for short-term (7-day) pulmonary rehabilitation in patients with AECOPD, producing superior improvements in dyspnea, health status, exercise capacity, and psychological well-being compared with conventional breathing training methods. The device’s automated personalised settings and real-time feedback provide a promising technical approach for optimising pulmonary rehabilitation during hospitalisation. Nevertheless, the scope of the evidence from this study must be clearly defined. The findings are limited to a supervised inpatient setting and a 7-day intervention period; they do not demonstrate the feasibility, effectiveness, or cost-effectiveness of the device in unsupervised environments such as outpatient clinics, community programmes, or home-based care. At the same time, conventional low-technology breathing exercises also provided clinically meaningful improvements during hospitalisation. Therefore, the device should not replace basic respiratory training but rather be considered a complementary or step-up option when resources permit. Author contributions All authors contributed to the study conception and design. Material preparation, respiratory training delivery, and manuscript drafting were performed by Yinan Xing. Data collation and statistical analysis were conducted by Guanhua Hou. Data organization, analysis support, and data collection were performed by Wei Li. Patient follow-up, outcome tracking, and data analysis were carried out by Zishuo Wang. The first draft of the manuscript was written by Yinan Xing and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript. Limin Zhao supervised the entire project and is the corresponding author. Funding The authors declare that no funds, grants, or other support were received during the preparation of this manuscript. Data availability Datasets generated and/or analysed during the current study may be obtained from the corresponding authors upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement Datasets generated and/or analysed during the current study may be obtained from the corresponding authors upon reasonable request.