This is a single-blind, prospective, randomized controlled trial (RCT) to determine the effectiveness, safety, and compliance of a DTx-based cardiopulmonary rehabilitation program. Random numbers were generated before the start of the enrollment by a researcher unrelated to this trial. The main researcher enrolled participants and assigned them to interventions according to the random number and the entry order. The grouping was blinded to patient identity. Outcomes were analyzed blinded to the study group. The reporting of this trial adheres to the CONSORT (Consolidated Standards of Reporting Trials) 2010 statement (Multimedia Appendix 1).

This trial was approved by the Research Ethics Board of the West China Hospital of Sichuan University (2022-1317) and registered at the Chinese Clinical Trial Register (ChiCTR2200064000). The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participating patients and they were allowed to withdraw the consent anytime during the study for any reason. Data were collected at West China Hospital. Patients’ data are anonymized. Each patient received a subsidy of RMB 800 (≈US $110).

The patients were recruited from West China Hospital. Patients who met all of the following criteria were included:

Patients who met any of the following criteria were excluded:

According to previous research from Messaggi-Sartor et al [46] on the improvement of VO₂ peak in postoperative exercise intervention for NSCLC, the sample size in this study was estimated to detect a difference of 2.13 mL/Kg/min in VO₂ peak, with a SD of 2.3 mL/Kg/min, a type I error rate α of 0.05, a power of 0.90 (1 – β), and an intraclass correlation coefficient of 0.10 for the multicenter design. It was determined that 45 participants would be required for an independent 2-sample t test comparison. Considering a 10% attrition rate, the intended total sample size was 50 (25 in the intervention group and 25 in the control group). PASS (version 15.0; NCSS) was used for sample size calculation.

Participants in the usual care group received routine postoperative rehabilitation guidance including condition monitoring, medication guidance, dietary guidance, oral exercise guidance, and psychological care. Besides the usual care, participants in the intervention group received exercise prescriptions with a video guide and real-time monitoring using the Recovery Plus Health app (Recovery Plus USA Inc), a software as medical device approved by the National Medical Products Administration for remote cardiac rehab and prehab. They were instructed to download the app for patients (Recovery Plus Inc; Multimedia Appendix 2), which was wirelessly connected to a chest-worn HR band (Recovery Plus Inc) to monitor exercise frequency, intensity, time, volume, and progression. These participants were shown how to operate the app and the equipment and were required to adhere to the prescribed exercises available on the app, wear the HR band during exercise, and provide feedback about the specific exercise after each rehabilitation session. The exercise prescriptions were generated according to the result of the cardiopulmonary exercise testing (CPET) and patients’ previous exercise habits. The prescription principle was based on suggestions for patients with cancer in American College of Sports Medicine’s guidelines [47] and Expert Consensus on Exercise Rehabilitation for Chinese Cancer Patients Centered on Functional Impairments [48]. Generally, aerobic exercise combining elastic band-based resistance exercises was recommended with a frequency of 3-5 d/wk and a duration of 90-150 min/wk. For each participant, the prescription was dynamically changed and highly personalized. Patient-reported, disease-related symptoms were considered during action selection. The effective exercise duration was recorded when patients’ HR reached the target HR zone, representing the duration under the target intensity. The lower number of the target HR zone was calculated as HR_resting + (HR_max – HR_resting) × 40%. The higher number of the target HR zone was calculated as HR_resting + (HR_max – HR_resting) × 60%. The prescription used a gradual progression and automatic adjustment in duration and action intensity based on the exercise data and feedback. All the adjustments automatically generated by the system were approved by the physician. The HCP reviewed the exercise records and feedback information via the Recovery Plus Health app for HCPs (Multimedia Appendix 3) weekly and made necessary modifications to the exercise regimen, while the app automatically recorded compliance with the exercise prescription. Regardless of the group, all participants were instructed to discontinue exercising and promptly report to their clinician or exercise professional if any cardiac-related symptoms occurred. Moreover, HCPs were available 24/7 to receive and respond to alerts from the participants. The intervention began 1-2 months after the surgery. Figure 1 illustrates the operational diagram of the AI-driven exercise prescription system.

Outcome measurements included cardiopulmonary fitness (VO₂ peak; primary outcome); lung function (forced expiratory volume in 1 second and diffusing capacity of the lungs for carbon monoxide); cardiac function (tricuspid annular plane systolic excursion and ejection fraction); cytokeratin 19 fragmentation antigen; safety; compliance (completion rate, exercise times per week, effective exercise duration, and prescription adherence rate); and scales assessing symptoms (MD Anderson Symptom Inventory for Patients With Lung Cancer [49,50]), psychology (Hospital Anxiety and Depression Scale [51]), sleep (Pittsburgh Sleep Quality Index [52]), fatigue (Multidimensional Fatigue Symptom Inventory—Short Form [53,54]), and QOL (European Organization for Research and Treatment of Cancer Quality of Life Questionnaire-Core 30 [55]). All the scales have been studied for validity and reliability. The assessment was conducted at baseline (1-2 months after surgery) and 5 months after the intervention, respectively. All the examinations were performed at West China Hospital and the scales were collected via the web. Exercise data including frequency, duration, HR, and subjective feedback were recorded by the app.

Descriptive statistics are presented as frequency distributions, means, and SD. Fisher exact test was used for unordered qualitative data. Independent 2-sample t tests were used for normally distributed continuous quantitative data with equal variance. The Mann-Whitney U test was used for ordered qualitative data and nonnormally distributed or heteroscedastic continuous quantitative data. The distribution of data was analyzed by the Shapiro-Wilk test. The equality of variances was tested using the Levene test. Statistical significance was established at a 2-tailed P value of less than .05. All analyses were conducted using an intention-to-treat approach.

From September 2022 to September 2023, a total of 47 patients in West China Hospital were enrolled in the study, with 40 (85%) individuals completing the trial (22 in the telerehabilitation group and 18 in the usual care group). Seven participants withdrew from the trial. One participant in the telerehabilitation group was diagnosed with asthma and the other 6 withdrew their consent because they were unable to come back to the hospital for examination on time due to living far away (Figure 2). The baseline data of participants who completed the study are presented in Table 1. The adherence rate of the telerehabilitation group was 88% (22/25). Patients in this group exercised 3.33 times a week on average, with an average exercise duration of 151.4 min/wk. The average effective exercise duration (reaching the prescribed target HR zone) was 92.25 min/wk. The mean prescription adherence rate (effective exercise duration to the lower limit of prescription duration) was 101.2%.

Four participants did not receive a CPET after the intervention and another 4 participants did not complete the scales on time. A total of 36 participants accepted both baseline and postintervention CPETs, with 20 individuals in the telerehabilitation group and 16 in the usual care group. Additionally, 36 participants responded to the baseline and postintervention scales, with 20 individuals in the telerehabilitation group and 16 in the usual care group. The cardiac telerehabilitation was associated with higher improvement of VO₂ peak (3.66, SD 3.23 mL/Kg/min vs 1.09, SD 3.23 mL/Kg/min; P=.02) and global health status or QOL (16.25, SD 23.02 vs 1.04, SD 13.90; P=.03) compared with usual care. Better alleviation of affective interference (relationships with others, enjoyment of life, and emotions; (–0.88, SD 1.50 vs 0.21, SD 1.22; P=.048), fatigue (–8.89, SD 15.96 vs 1.39, SD 12.09; P=.02), anxiety (–0.31, SD 0.44 vs –0.05, SD 0.29; P=.048), and daytime dysfunction (–0.55, SD 0.69 vs 0.00, SD 0.52; P=.02) were also observed in the telerehabilitation group (Table 2). No significant differences in other efficacy indicators were observed between the 2 groups (all P>.05). No exercise-related adverse events were identified during the intervention period.

This RCT shows the effects of DTx-based telerehabilitation for NSCLC survivors. Both cardiopulmonary fitness and QOL were significantly improved by telerehabilitation compared with usual care. Previous research has demonstrated that exercise-based rehabilitation can enhance cardiopulmonary fitness across a diverse range of disease conditions. A similar RCT in early-stage NSCLC survivors has shown that 8-week aerobic exercise combining high-intensity respiratory muscle training improved VO₂ peak (2.13 mL/Kg/min). However, in this CBCR program, only 69% (11/16) participants persisted in the exercise group with a compliance (session completion rate) of 80%, and no significant QOL improvement was observed [46]. Similar low adherence has been reported in many rehabilitation studies for NSCLC [56,57], especially in advanced NSCLC survivors after surgery [58,59]. As for the impact of exercise interventions on the QOL, conflicting findings have been reported in lung cancer survivors [60]. A meta-analysis involving 34 RCTs demonstrated that unsupervised HBCR did not improve QOL in cancer survivors compared with supervised programs [61]. Patients in unsupervised programs often fail to reach the recommended exercise dosage, which is the main reason for the observed absence of QOL improvement. Maintaining high adherence to exercise routines is a crucial approach to ensure the benefits [60]. Integrating wearable devices into HBCR presents a flexible solution that overcomes the limitations of traditional CBCR, offering patients a more adaptable and convenient way to access rehabilitation programs. In this study, 88% (22/25) of the patients in the telerehabilitation group completed the trial with an impressive average prescription compliance rate of 101.2%, indicating a strong engagement and adherence to the prescribed exercise regimen. DTx enables patients to actively participate in their health management by delivering ongoing feedback and encouragement, motivating them to maintain regular physical activity.

Exercise type affects the outcome of rehabilitation. Aerobic training is particularly effective in improving VO₂ peak and associated cardiopulmonary variables while resistance training primarily focuses on enhancing overall muscular health. An RCT reported that aerobic training rather than resistance training improved VO₂ peak in lung cancer survivors with poor cardiorespiratory fitness [62]. However, a combination of aerobic exercise and resistance exercise can enhance various health outcomes commonly affected by cancer [63]. Another factor that affects the outcome of rehabilitation is the intensity. High-intensity exercise, characterized by a metabolic equivalent of task greater than 6, is defined as an activity that induces sweating and elevates heart and respiratory rates. It has been shown to decrease cancer progression and enhance survival rates in breast, colon, and prostate cancers, with emerging evidence suggesting similar benefits in lung cancer [64]. The exercise prescription in this program combined aerobic and resistance exercises and displayed real-time HR on the guidance interface to help patients maintain an effective exercise intensity.

In addition to good compliance and effectiveness, the following 2 features of this system make it promising for clinical applications. First, the tag-based action recommendation system allows for a high degree of individualization. This is reflected not only in action selection based on patient symptoms and duration settings according to previous exercise habits but also in the ability to adjust action intensity and duration dynamically based on exercise data and feedback. Second, this system enables clinicians to manage a large number of patients simultaneously via the web, viewing patient exercise data and feedback intuitively, thus achieving high management efficiency. These characteristics are difficult to achieve with existing CBCR systems.

This study has certain limitations that warrant consideration. First, the relatively short duration of the intervention, which was limited to 5 months, did not allow us to observe long-term outcomes such as prognosis. Implementing the intervention through an app introduces a learning curve for patients. The novelty of digital therapeutics can also be a limitation. Therefore, a longer follow-up period is needed to understand the sustained effects on disease progression and participants’ awareness, knowledge, attitudes, intentions to change, help-seeking behaviors, and behavior change, providing a fuller picture of the long-term benefits of HBCR. In addition, a rehabilitation regimen that encompasses a broader time frame, including the perioperative period, could potentially yield greater benefits. Second, the focus of this study was on exercise prescription, which, while a central aspect of cardiac rehabilitation, is only one part of a complex intervention. Comprehensive cardiac rehabilitation programs should also include nutrition guidance, psychological support, medication management, smoking cessation efforts, patient education, and diligent monitoring of key health indicators. Incorporating these multifaceted components could significantly enhance the program’s effectiveness and the overall wellness of lung cancer survivors. Third, this study is centered on patients with early-stage, postoperative NSCLC due to their extended survival expectancy, which correlates with a heightened risk of CVDs in the future. They can tolerate most of the exercise prescriptions and intensities, which positions them as the ideal candidates for telerehabilitation following lung surgery. However, for patients with advanced NSCLC, additional treatments such as radiotherapy and chemotherapy can lead to more significant side effects and cardiopulmonary damage. It is essential to conduct further studies to investigate telerehabilitation modalities and their effectiveness for these patients.

This study aimed to expand the application of digital health care in cardiopulmonary rehabilitation after lung cancer surgery. The 5-month, DTx-based telerehabilitation improved cardiorespiratory fitness in lung cancer survivors with good compliance and safety. Patients receiving telerehabilitation also reported improved QOL with reduced levels of fatigue, anxiety, and daytime dysfunction. The findings indicate that incorporating DTx-based cardiac rehabilitation into postoperative care may benefit early-stage NSCLC survivors. More work should be done to address the long-term impact on prognosis.