This prospective, parallel, 2-armed, single-blinded, single-center RCT was prospectively registered at ClinicalTrials.gov (NCT05197010) using CONSORT-EHEALTH (Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth) guidelines (Checklist 1). The registered protocol comprised 2 separate RCTs involving different patient populations—chronic low back pain and knee osteoarthritis—each using a tailored version of the mobile app. This study analyzed the RCT conducted in the knee osteoarthritis cohort.

The study participants were recruited from the Department of Physical and Rehabilitation Medicine at Samsung Medical Center, Seoul, Republic of Korea between October 2021 and December 2021. Study information was posted on the hospital’s bulletin boards and provided to patients clinically diagnosed with knee osteoarthritis who visited the outpatient rehabilitation clinic. Individuals who expressed interest received detailed information about the study, and written informed consent was obtained prior to screening. Eligibility assessments were conducted only for participants who provided consent. Inclusion criteria were: (1) age ≥60 years; (2) knee pain persisting for ≥3 months; (3) overall knee pain severity ≥4 on 11-point Numeric Rating Scale (NRS); (4) score ≥23 points on the Korean version of the Montreal Cognitive Assessment [22], indicating ability to comprehend exercise program; and (5) radiographic evidence of osteoarthritis with Kellgren-Lawrence (K-L) grade ≥2 [23]. K-L grade was assessed by a musculoskeletal specialist (JGD) with over 10 years of clinical experience in musculoskeletal imaging. For patients with bilateral knee osteoarthritis, the most symptomatic knee—defined as the knee with greater pain as reported by the patient—was used for eligibility screening. The same knee was consistently evaluated at baseline and follow-up outcome assessments. Participants who met the following criteria were excluded: (1) a history of knee surgery, (2) a history of intra-articular or periarticular knee injections within the past 3 months, (3) a history of systemic inflammatory disease, (4) a history of polyneuropathy, (5) a history of hemorrhagic or ischemic cerebral infarction, (6) a history of heart failure or chronic obstructive pulmonary disease affecting walking or daily activities, and (7) other medical conditions unsuitable for self-directed home exercise (eg, severe anemia and uncontrolled diabetes mellitus).

Data on demographics, anthropometric, medical history, and social history were obtained from the participants prior to randomization.

Eligible participants were randomized to the intervention or control group in a 2:1 ratio using block randomization with a block size of 6. A computer-generated randomization list was prepared at the study outset and was securely concealed in opaque envelopes. The randomization process was conducted by an independent researcher who was not involved in the outcome assessments, ensuring blinding to the treatment allocation and randomization. To further minimize bias, statistical analyses were performed by an independent analyst who was not involved in the randomization process, participant allocation, or data collection. Although the study was single-blinded, with participants aware of the use of a mobile app, the study hypothesis was not disclosed to avoid potential bias in patient-reported outcomes.

The intervention group participants underwent a 6-week self-exercise program delivered via a mobile app–based self-exercise program (Multimedia Appendix 1). During their initial visit, participants were instructed to install the app on their mobile devices and received comprehensive training on its use. The app guided users through a daily 30-minute self-exercise protocol, comprising stretching, strengthening, functional, and cooldown exercises (Multimedia Appendix 1). The exercise program was tailored by physiotherapists based on the patient’s pain scale, functional status, and overall condition, ensuring a personalized approach to rehabilitation. To monitor the progress and adjust the intensity and content of the exercise program, the app incorporated a patient-reporting system. Before each daily exercise session, the participants quantified their pain intensity using a 5-point Likert scale. After completion, they again rated their perceived exertion using a 5-point Likert scale. On a weekly basis, the participants completed the KOOS-PS within the app. Accordingly, the app provided immediate visual feedback to the participants, and participants could check their weekly statistics for exercise achievement and pain levels within the app (Figure 1). Furthermore, these data were transmitted to the supervising physiotherapist and physiatrist for remote monitoring and exercise program modifications. The exercise program was adjusted weekly to 1 of the 3 intensity levels (ie, high, moderate, and mild) based on patient-reported outcomes (Figure 2). Owing to this approach, the app delivered a personalized self-exercise program throughout the 6-week intervention period.

Participants in the control group engaged in a 6-week self-exercise program, comprising daily 30-minute sessions without using a mobile app. During their initial visit, participants received a single session of individualized exercise education from a qualified physiotherapist. The exercise program components were the same as those in the intervention group, namely, stretching, strengthening and functional, and cooldown exercises. Furthermore, the participants were provided with a paper brochure detailing the prescribed exercise program.

No additional treatments for knee osteoarthritis, such as intra-articular injections, physical modalities, or surgical procedures, were initiated or prescribed during the 6-week intervention period in either group. Participants who were taking analgesics or nonsteroidal anti-inflammatory drugs for knee osteoarthritis at baseline were instructed to maintain their usual regimen without changes throughout the study period. Use of medications for other comorbid conditions was permitted. Any changes in additional treatments for knee osteoarthritis were monitored and documented at the 6-week follow-up visit.

To evaluate feasibility, the retention rate was calculated as the proportion of participants who completed the 6-week follow-up assessment relative to those randomized. For participants who withdrew, the number and reasons for dropout were recorded. In addition, participants in the intervention group completed a survey to evaluate the mobile app–guided self-exercise program at the 6-week follow-up. The survey collected data on self-reported exercise frequency, participants’ expectations for achieving outcomes through mobile app use, and satisfaction levels with mobile app–guided self-exercise programs, as assessed using the North American Spine Society Patient Satisfaction Index [24]. For the safety outcome, any adverse events that occurred during the intervention period were recorded throughout the study.

According to the registered protocol of the main study (NCT05197010), a total of 60 participants (30 with chronic low back pain and 30 with knee osteoarthritis) were planned for enrollment. The sample size for each population was calculated based on recommendations from the literature on conducting pilot studies [34-36]. Assuming an expected effect size of 0.5 [6] and aiming for 90% power at a 2-sided significance level of 0.05, a total of 28 to 30 participants (14‐15 per group) were necessary for the pilot trial [35,36]. To facilitate recruitment and encourage participation, a 2:1 allocation ratio was applied, resulting in 20 and 10 participants in the intervention and control groups, respectively. This paper reports data from the knee osteoarthritis cohort only, which was one of the 2 independent randomized controlled trials (chronic low back pain and knee osteoarthritis) included in the registered main protocol (NCT05197010).

All analyses were performed using R (version 4.2.1; R Foundation for Statistical Computing). The primary analysis followed a modified intention-to-treat principle, including all randomized participants who received at least 1 session of the allocated intervention and had postbaseline outcome data available. This approach was adopted because excluding participants who did not initiate the intervention enables estimation of the treatment effect within the subpopulation who would have initiated treatment [37]. The Shapiro-Wilk test was used to assess data normality. Between-group comparisons of baseline characteristics and outcome measures at baseline and at the 6-week follow-up were performed using the Fisher exact test for categorical variables. For continuous variables, either the 2-tailed independent t tests or the Wilcoxon rank-sum test was used, based on the normality of the data.

The preliminary efficacy analysis evaluated the between-group differences in changes from baseline to 6-week follow-up. Between-group differences in changes (calculated as intervention group minus control group) were analyzed using independent t tests with mean differences and 95% CI when normality assumptions were met. Otherwise, the Wilcoxon rank-sum test was performed, and the median differences with 95% CI were estimated using the Hodges-Lehmann method. For within-group comparisons (calculated as 6-week follow-up minus baseline), a paired t test or a Wilcoxon signed-rank test was performed based on the normality of the data.

Statistical significance was set at P<.05 (2-tailed), and cases with missing data were excluded from the analysis.

The study protocol was reviewed and approved by the Institutional Review Board of Samsung Medical Center (2021-04-004) and was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to enrollment. To ensure privacy and confidentiality, all collected data were deidentified. Participants received compensation for transportation costs of 50,000 KRW (approximately US $40) per visit, totaling 100,000 KRW (approximately US $80) for two visits, based on the exchange rate at the time of the study.

During the 3-month recruitment period, a total of 29 participants provided informed consent and underwent an eligibility assessment, and none met the exclusion criteria. Specific reasons for declining participation among individuals who refused screening were not recorded. Therefore, the participant flow diagram begins with the number of participants who provided consent and were screened (Figure 3). In this study, 29 participants (mean age 70.4, SD 6.6 y; n=24, 83% female participants) were randomized in a 2:1 ratio to the intervention and control groups. Three participants in the intervention group withdrew before initiating the assigned intervention. Two withdrew due to poor general condition related to systemic illness, and one withdrew because of travel abroad (Figure 3). These 3 participants were excluded from subsequent analyses. A total of 26 participants (n=16, 62% in the intervention group and n=10, 38% in the control group) completed the 6-week follow-up, yielding an overall retention rate of 90% (26/29; 16/19, 84% for the intervention group and 10/10, 100% for the control group). No adverse events were reported in either group during the study period. At baseline, there were no significant differences between the 2 groups in any clinical or demographic characteristics, including K-L grade (Tables 1 and 2).

All 16 participants in the intervention group completed the mobile app–related survey. During the intervention period, adherence to the mobile app–based exercise program was high: 69% (n=11) of participants reported exercising ≥5 days per week, and 38% (n=6) reported daily exercise. The remaining participants reported exercising at least twice weekly, with 13% (n=2) exercising 2 to 3 days per week and 19% (n=3) exercising 3 to 5 days per week.

Regarding participants’ goals with our mobile app, 50% (n=8) of participants aimed for pain relief, 44% (n=7) sought improvement in muscle strength, and 6% (n=1) expected enhanced quality of life. Overall, 88% (n=14) reported high satisfaction (scores 1‐2 on the North American Spine Society Patient Satisfaction Index). None of the participants in either group changed their medication regimen or received additional treatments for knee osteoarthritis during the study period.

The WOMAC total score exhibited a significantly greater reduction in the intervention group compared with the control group from baseline to the 6-week follow-up (between-group median difference in change –12.00, 95% CI –28.00 to –2.00; P=.02), indicating a significantly greater improvement in the intervention group (Tables 3 and 4). Within-group analysis revealed a significant reduction in the WOMAC total score in the intervention group (median change –11.00, 95% CI –23.00 to −2.50; P=.01), whereas no significant changes were observed in the control group (median change 4.45, 95% CI –16.00 to 20.00; P=.39; Table 5).

Among the WOMAC subscales, the physical function score exhibited significantly greater improvement in the intervention group than in the control group (between-group median difference in change –10.00, 95% CI –25.00 to –4.00; P=.01; Table 4). Within-group analysis revealed significant reductions in the stiffness and physical function scores in the intervention group (mean change in stiffness –1.19, 95% CI –2.16 to –0.21; P=.02; median change in physical function –10.50, 95% CI –19.50 to –2.00; P=.01), whereas no significant changes were observed in the control group (P>.05; Table 5).

A between-group comparison of changes from baseline to the 6-week follow-up revealed that the improvements in NRS pain were statistically greater in the intervention group than in the control group (between-group mean difference in change –2.02, 95% CI –3.94 to –0.11; P=.04). No significant differences in changes in the remaining secondary clinical outcome measures were observed between the 2 groups, including KOOS-PS, SF-36, K-GDS, TUG test, 30-CST, handgrip strength, knee extensor strength, and thickness of the quadriceps femoris muscle (P>.05 for all; Tables 3 and 4).

Significant within-group improvements were observed in the intervention group at the 6-week follow-up, including reductions in NRS pain (mean change –2.12, 95% CI –3.13 to –1.11; P<.001), KOOS-PS (median change –2.50, 95% CI –4.00 to 0.00; P=.046), and TUG test for the right side (median change –1.10 s, 95% CI –2.08 to –0.47; P=.006). Furthermore, significant increases were noted in knee extensor strength (mean change 6.74, 95% CI 3.33-10.15; P<.001) and left vastus intermedius thickness (median change 0.17 cm, 95% CI 0.02-0.27; P=.03; Table 5). However, the control group exhibited no significant improvements in secondary outcomes, except for the knee extensor strength, which increased by 6.12 kgf (95% CI 2.38-9.86; P=.005).

This pilot randomized controlled trial demonstrated that a mobile app–based personalized self-exercise program was feasible, acceptable, and safe for older patients with knee osteoarthritis. The study achieved a high retention rate in both groups (16/19, 84% for the intervention group and 10/10, 100% for the control group). Although 3 participants in the intervention group withdrew, all reasons for withdrawal were unrelated to the mobile app or study procedures. These findings indicate good participant compliance and minimal burden over the 6-week intervention period. High adherence (nearly 70% performing exercise ≥5 d/wk) and strong participant satisfaction (14/16, 88% reporting high satisfaction) of the intervention group indicate that the program’s design—featuring personalized exercise adjustment and visualized feedback—was well tolerated and engaging for older users.

Although this pilot study was not powered to detect definitive efficacy outcomes, exploratory analyses revealed improvements in pain and physical function among participants using the mobile app compared with traditional paper-based methods. The observed effect sizes for physical function (WOMAC total: –0.63, 95% CI –1.11 to –0.15) and pain severity (NRS: −1.10, 95% CI –1.74 to –0.46) indicate the potential effectiveness of our mobile app. Furthermore, these clinical outcomes aligned with the patients’ needs, with 94% of the participants in the intervention group anticipating improvements in pain or muscle strength. Taken together, the favorable feasibility outcomes and preliminary clinical trends support the practicality and promise of conducting a larger, adequately powered randomized controlled trial to verify clinical effectiveness.

A 2024 Cochrane systematic review reported that exercise interventions improve clinical outcomes, including immediate pain relief, physical function, and quality of life (absolute effect sizes: 0.72 for physical function; 0.71 for pain), compared with no treatment, usual care, or limited education [7]. Despite the established efficacy of exercise interventions for knee osteoarthritis, the limited use of digital delivery methods in these trials highlights a crucial gap in clinical studies. Only a small fraction (6%) of the reviewed trials incorporated digital delivery methods for exercise interventions, with most relying on in-person methods [7].

Reportedly, digital tools incorporating personalized feedback and monitoring improve pain management and functional outcomes in chronic musculoskeletal conditions [38-40]. For knee osteoarthritis, several studies have explored the use of mobile apps for home-based exercise or self-management in knee osteoarthritis [10,13-15,17-19]. Although 1 study used a decision-tree algorithm to determine the disease stage and recommend an appropriate exercise program for knee osteoarthritis at baseline [17] and some mobile apps tracked metrics such as exercise duration [10] and step counts [13], none monitored patient symptoms within the app consistently. The mobile app used in this study addressed this gap by combining the ongoing patient-reported symptom monitoring with responsive feedback and exercise modifications, representing a distinct advancement over existing mobile app interventions.

In a prior study using a paper-based home exercise program for knee osteoarthritis, the mean adherence rate at 6 weeks was 60.2% (SD 33.1%) [10]. In our study, 69% (11/16) of participants in the intervention group exercised 5 or more days per week, indicating a potential for improved exercise adherence with the use of a mobile app. By combining established exercise benefits with user-friendly technology, this mobile approach may enhance adherence and engagement, thereby supporting improved pain control, physical function, and overall quality of life in older patients with knee osteoarthritis.

A key strength of our app is the high retention and adherence rate, which is particularly significant considering the documented challenges in older populations [41]. The 2024 Cochrane systematic review noted a decline in the sustained efficacy of exercise at 12 months (absolute effect sizes reduced to 0.32 for pain and physical function) [7]. These findings highlight the importance of adherence in maintaining the long-term benefits of exercise interventions. The success of the app in maintaining adherence can be attributed to three key features. First, the automated pop-up alarms delivered via participants’ mobile phones are active reminders to encourage consistent participation in the exercise program. Second, the visualized statistical tracking features and weekly patient-reported symptom monitoring system of the app enable participants to continuously assess their progress, providing tangible feedback on their exercise achievements. Third, the dynamic modification of exercise intensity based on individual progression would maintain appropriate challenge levels while accommodating participants’ changing capabilities. These interactive features of digital technology may address common barriers to exercise adherence, as demonstrated by high participation rates and significant improvements in subjective and objective outcomes in the intervention group. Herein, the superior outcomes observed in participants who completed the mobile app–based self-exercise program compared with those who performed paper-based self-exercises may be attributed to the dynamic adjustments and real-time feedback of the program; these findings suggest that a user-friendly, personalized mobile app–based self-exercise program may achieve therapeutic benefits superior to traditional brochure-guided self-exercises and comparable to in-person interventions while enhancing accessibility and convenience for patients.

In addition to the clinical outcomes, this study identified possible physiological changes associated with the mobile app–based exercise program. In particular, an increase in the left vastus intermedius muscle thickness (median change 0.17 cm, 95% CI 0.02 to 0.27; P=.03) of the intervention group may indicate a trend toward muscle mass gain following participation in the program. This finding is noteworthy because reduced lower limb muscle strength and muscle mass are well-documented in symptomatic knee osteoarthritis [42] and risk factors for falls among older populations [43]. Patients with knee osteoarthritis are particularly susceptible to recurrent falls [44], and fall-related injuries are a leading cause of morbidity and mortality among the older population [41,45]. Thus, exercise interventions targeting muscle strength and mass are crucial for mitigating fall risk. In our study, within-group analysis revealed significant increases in knee extensor strength in the intervention and control groups. However, improvements in the TUG test, a key indicator of fall risk, were exclusively observed in the intervention group (median change –1.10 s, 95% CI –2.08 to −0.47; P=.006). This discrepancy may be attributed to the concurrent increase in the vastus intermedius muscle mass observed in the intervention group, underscoring the potential of mobile app–based self-exercise programs in reducing the risk of falls in patients with knee osteoarthritis.

Nevertheless, this study was preliminary, and ultrasound-based muscle thickness measurements are subject to measurement variability [46]. The increase observed only in the left vastus intermedius may represent measurement variation or asymmetrical limb use rather than a consistent hypertrophic response of the muscle. Therefore, these findings should be interpreted as exploratory, and confirmation in a larger, well-controlled trial is warranted.

The relatively small sample size and the high proportion of female participants in the study sample (24/29, 83%) may limit the generalizability of the findings. Another limitation of this study is that exercise adherence was assessed only in the mobile app group, and direct comparisons with the paper-based control group were not conducted. Therefore, it remains unclear whether the higher adherence observed reflects the effect of the digital intervention itself or general participant motivation. In addition, the relatively short 6-week follow-up period may be insufficient to completely assess the sustainability of the observed improvements. Thus, future studies with larger sample sizes and longer follow-up periods are warranted to validate these findings and evaluate their long-term clinical impact.

This pilot randomized controlled trial demonstrated that a mobile app–based self-exercise program is feasible, acceptable, and safe for older patients with knee osteoarthritis. Over the 6-week intervention, participants in the mobile app group showed high adherence, strong satisfaction, and favorable trends in pain reduction and physical function compared with those using a paper-based program. By combining the proven benefits of exercise with the accessibility of digital technology, this approach shows promise for enhancing engagement and overcoming common barriers to exercise adherence. Future large-scale randomized controlled trials are warranted to confirm these preliminary findings and to evaluate the long-term effectiveness and clinical integration of mobile health interventions for knee osteoarthritis management.