We enrolled 50 patients with CAD who were eligible for outpatient CR (phase III) for the secondary prevention of CAD at Juntendo University Hospital between April 30, 2022, and January 21, 2023. The exclusion criteria were as follows: (1) implantable medical device users such as patients with cardiac pacemakers and defibrillators; (2) patients diagnosed with acute myocardial infarction, uncontrolled angina pectoris not stabilized by medical treatment, uncontrolled arrhythmias causing subjective symptoms or hemodynamic abnormalities, uncontrolled heart failure, symptomatic severe aortic stenosis, or a psychiatric disorder impeding communication; and (3) participants deemed unsuitable for this study by the principal investigator and research physician.

This was a randomized, open-label, comparative, parallel-group, single-center interventional study. This study followed the CONSORT (Consolidated Standards of Reporting Trials) reporting guidelines. In this study, patients were randomly divided into a wearable device plus OLC group and a control wearable device (CON) group, stratified by age (<60 years and ≥60 years) and sex (male and female), using a stratified permuted block design as the allocation method. The smartphones used by the patients were personal devices, and the wearable device was a Fitbit Sense. Patient data were collected using a wearable device linked to a smartphone for real-time monitoring of step count, activity level, and heart rate (Recoval; for detailed information about the product, please see [16]). A proprietary algorithm converted the acceleration signals from a wearable device into step counts and metabolic-equivalent tasks (METs). The intervention lasted for 12 weeks, and all patients underwent a cardiopulmonary exercise test (CPET)-based home exercise program. This program aims to engage in exercise at anaerobic threshold (AT) heart rate and a Borg scale of 12‐13, for at least 30 minutes per day, 5 days a week. Additionally, the patients in the OLC group were provided with exercise guidance based on their activity levels recorded by their wearable device. Guidance was provided via weekly text messages and monthly videoconferences. The participants in the CON group wore only the wearable device and did not receive any additional instructions (Figure 1).

The Ethics Committee of the Juntendo Clinical Research and Trial Centre approved this study (IRB number E21-0353). Patients were enrolled by outpatient primary care physicians who were not involved in the analysis, and external clinical research coordinators allocated them into the 2 groups. All patients provided written informed consent to participate in this study and had the option to opt out of this study at their discretion. The collected data were deidentified and managed accordingly. Additionally, no compensation was provided to study participants.

The sample size calculation was based on anticipated changes in peak VO₂ between the OLC and CON groups, as reportedly previously by comparable studies. We assumed a mean improvement in peak VO₂ of 2.0 mL/kg/min in the OLC group and 1.0 mL/kg/min in the CON group, estimating this as approximately 80% of the peak VO₂ increase observed in comparable studies [24]. To determine the sample size, we used a pooled SD of 1.1 mL/kg/min, a 2-sided significance level of 5%, and a power of 80%, which led to an initial requirement of 22 participants per group. To account for an anticipated 10% dropout rate, we adjusted the final sample size to 25 participants per group.

Categorical data are presented as numbers and percentages and were compared using the chi-square test. Continuous variables are expressed as mean (SD) or median (IQR). Between-group comparisons were performed using Student t test, while within-group comparisons were performed using a paired t test. A P value <.05 was considered statistically significant. Effect sizes for the group×time interaction were reported using Cohen f, interpreted as small (f≥0.10), medium (f≥0.25), and large (f≥0.40). Missing values were not imputed, and multiplicity was not considered. In this pilot trial, the outcome assessors were blinded to group allocations. All statistical analyses were performed using R (version 4.4.2; R Foundation).

A total of 83 patients with CAD were enrolled in this study. Of these, 6 patients were excluded because they did not have compatible smartphones, 23 were excluded due to lack of consent, 1 was excluded due to uncontrolled angina, and 3 patients were excluded due to uncontrolled arrhythmias. As a result, 50 patients with CAD were finally analyzed and classified into 2 groups, OLC and CON, with randomization yielding 25 participants per group. One patient in the OLC group dropped out during the follow-up period because of lower limb muscle strain. No major adverse cardiac events occurred during this study’s period; these were defined as a composite of cardiovascular death, nonfatal myocardial infarction, nonfatal stroke, and admission for heart failure. Finally, baseline data and data at 12 weeks following intervention from a total of 49 patients were included in the analysis (Figure 2). Table 1 presents the patient backgrounds. The mean age (63.8, SD 6.4 vs 62.6, SD 7.9 years) and sex distribution (n=23, 95.8% vs n=22, 92% male) between the OLC and CON groups were similar. No significant differences were observed between the 2 groups regarding baseline characteristics and coronary risk factors (all P>.05). However, the use of β-blockers was significantly higher in the OLC group (n=19, 76%) than in the CON group (n=11, 45.8%; P=.03). There were no significant differences in baseline blood data between the 2 groups.

A baseline comparison showed no significant differences between the 2 groups regarding CPET parameters, including peak VO₂ (21.2, SD 3.5 vs 20, SD 5.9; P=.39) and AT VO₂ (13.1, SD 2.3 vs 12.4, SD 3.9; P=.40). Similarly, no significant between-group differences were observed concerning mental status measured by STAI and in HR-QOL measured by SF-36 (Table S1 in Multimedia Appendix 2).

After 12 weeks of intervention, both groups exhibited changes in their health metrics (Table 2). In the OLC group, there was a slight reduction in BMI (P=.003). Both groups exhibited a decrease in appendicular skeletal muscle mass and a slight increase in body fat percentage; however, these changes were not significant. In terms of blood data, high-density lipoprotein cholesterol levels significantly increased in both groups (P=.04).

Over a period of 12 weeks, changes in CPET parameters and mental status outcomes were observed in both groups (Table 3). In the OLC group, there was a significant increase in peak VO₂ with a change of 9 (SD 10.3%; P<.001), AT VO₂ with a change of 11.8 (SD 16%; P=.001), peak VO₂/HR with a median change of 0.5 (range −0.2 to 0.7, P=.04), and ΔVO₂/Δload with a mean range 1 (SD 0.8 range 0.4 to 3, P=.001). Similarly, in the CON group, significant improvements were noted in peak VO₂ with a mean change of 8.1 (SD 14%; P=.008), AT VO₂ with a mean change of 12.9 (SD 19.3%; P=.002), and ΔVO₂/Δload with a mean change of 2.0 (range 0.1 to 3.2, P=.001). The changes in peak VO₂, AT VO₂, peak VO₂/HR, and ΔVO₂/Δload over 12 weeks did not show a significant difference between the 2 groups. Concerning mental status, the OLC group showed a significant increase in body pain scores. Conversely, the CON group did not show significant changes in body pain levels.

The parameters measured by the wearable device are shown in Figures 3-6. Figure 3 displays the average number of daily steps taken each week. Although the increase in steps in the OLC group was not significant, a notable increase in the number of steps was observed. After 11 weeks, there was a significant increase in the number of steps compared with that in the CON group (10,785 steps vs 8463 steps; P=.03). Figure 4 displays the average daily travel distance per week. The distance covered was significantly longer for OLC group, especially in the latter period, compared with that in the CON group (8 km vs 6.5 km; P=.003). Of activity level, the daily duration of high activity was significantly longer for the OLC group after 9 weeks than that for the CON group (37 min vs 21.7 min; P=.03). However, there was no significant difference in the duration of moderate activity between the 2 groups (Figures 5 and 6).

We investigated the effects of remote CR using wearable devices with and without OLC in patients with CAD. The major findings of this study were as follows: (1) using wearable devices and providing a home exercise program based on CPET for 3 months significantly improved peak VO₂ and AT VO₂; (2) the OLC group demonstrated significantly higher physical activity (daily distance and daily duration of high activity) in the latter half of the intervention than did the CON group; and (3) during the research period, there were no dropouts owing to the use of wearable device or OLC.

This study had some limitations. First, there was a potential selection bias in the cases enrolled in this study, as it may have included many individuals with high health consciousness and tolerance for wearable devices and desire for OLC. These factors may limit the generalizability of our findings to population groups with lower levels of health consciousness or technological proficiency. Second, this was a single-center study conducted primarily among Japanese patients, with a small sample size and a short follow-up period of 12 weeks. Regional differences and cultural factors may have influenced the results. Furthermore, the predominance of the male sex in this study’s cohort may also limit the applicability of the findings to females, older adults, or different ethnic groups. Additionally, varying levels of digital literacy and access to technology could impact the effectiveness of remote CR programs, both within Japan and internationally. Third, the power analysis indicates that this study is sufficiently powered to detect large effect size differences between the OLC group and CON group, specifically a Cohen d of 0.8 with a sample size of 50. These effect sizes are rather large and may be unrealistic for the intervention being studied, suggesting that this study is underpowered for detecting more moderate or smaller effect sizes. This study serves primarily as a proof of concept. Future interventional studies with larger sample sizes, longer follow-up durations, and more diverse populations are needed to confirm the findings and evaluate long-term adherence and sustained benefits of such programs.

Peak VO₂, measured using the CPET, represents the maximal ability of the body to transport and use oxygen. It is widely recognized as a definitive measure of exercise capacity [25-28]. In patients with CAD, peak VO₂ is associated with all-cause mortality and cardiovascular events [29-32]. AT VO₂, similar to peak VO₂, is considered a prognostic indicator [33,34]. It is recommended to prescribe exercise below the AT to establish a safe exercise tolerance range of intensity for patients [5]. In this study, by providing a home exercise program based on AT VO₂ and using wearable devices, we identified significant enhancements in both peak VO₂ and AT VO₂ over a 12-week intervention period, irrespective of the presence or absence of OLC. Additionally, we observed a significant improvement in ΔVO₂/Δload. This parameter serves as an indicator of oxygen transport to the peripheral muscles, reflecting the degree of increase in cardiac output during exercise [35]. It has been reported that ΔVO₂/Δload is an independent prognostic factor in patients with cardiovascular disease [36]. In previous studies, it has been reported that the use of physical activity trackers for more than 12 weeks improves peak VO₂ in patients with CAD, however there is no mention of CPET parameters other than peak VO₂, such as AT VO₂ or ΔVO₂/Δload [15,37]. Additionally, the patient group had already participated in a phase II CR program. Our study targeted patients in phase III and did not consider whether they had participated in the CR program, suggesting that a combined approach using wearable devices and a CPET-based home exercise regimen offers a promising strategy for remote CR in phase III patients with CAD, potentially leading to improved prognosis and enhanced exercise capacity. However, this study was a short-term intervention of 12 weeks, and determining its effects requires a long-term intervention and follow-up.

The World Health Organization guidelines recommend at least 150‐300 minutes of moderate physical activity weekly, at least 75‐150 minutes of vigorous physical activity weekly, or an equivalent combination of both [38]. It has been reported that moderate-to-vigorous-intensity activity and vigorous physical activity, as measured by wearable devices, have a significant negative correlation with the risk of all-cause mortality, cardiovascular death, and onset of heart failure [8,9,39]. In this study, the OLC group had a significantly increased daily distance travelled compared with that by the CON group in the latter period. Additionally, while there was no significant difference in the time spent moderately active (3‐6 METs), the time spent highly active (>6 METs) significantly increased in the OLC group during the latter half of the intervention. To date, no study has examined the significance of OLC using wearable devices. In this study, weekly OLC had a positive effect on daily activity, particularly vigorous activity, in the latter half of the remote CR program. One possible reason is that OLC may help maintain the motivation to continue exercise habits and provide guidance on appropriate exercise methods. Notably, 1 participant in the OLC group dropped out owing to calf muscle strain, and the OLC group experienced increased body pain as measured by HR-QOL. This suggests that an increase in activity may lead to musculoskeletal problems. To safely continue remote-CR, it is important to not only increase vigorous activity but also to enhance moderate activity. Patient education and OLC are crucial in this regard.

With the advancement of digital devices, including wearable devices and smartphones, it is anticipated that remote-CR will continue to evolve in the future [40]. Prior research on remote CR using these digital platforms has primarily focused on relatively healthy patients in stable condition. Nonetheless, reported dropout rates in studies with a duration under 12 weeks ranged 3%‐13%, even among patients who demonstrated a degree of digital device usage tolerance [41-43]. In this study, the patients were predominantly male (>90%), with an average age of 63.1 (SD 7.2) years, indicating that the cohort was less likely to be affected by the digital divide. Future research must prioritize the validation of remote CR in older and more vulnerable populations, while also assessing the efficacy and safety of interventions over extended periods.

Remote CR using wearable devices and a real-time monitoring system significantly improved exercise capacity in patients with CAD over a 12-week intervention period. Physical activity improved significantly with regular OLC. Future research should focus on diverse patient populations and longer intervention duration to validate these findings.