Development of video exercise-based mobile application to improve the clinical outcomes in patients with knee osteoarthritis: a randomized controlled trial Authors: Fatih Özden, Baki Umut Tuğay, Osman Çoban, Barış Ethem Süzek, Nazan Tuğay, Zübeyir Sarı, İrem Çetinkaya, Gülser Cinbaz, İsmet Tümtürk Published: 2025 ABSTRACT AbstractObjectiveThe purpose of this study was to develop a video exercise-based mobile application and investigate its effectiveness in terms of pain, function, expectation, and satisfaction in patients with knee osteoarthritis (OA).MethodsA randomized controlled trial was carried out with 52 individuals with knee OA. Participants were randomly allocated into two groups: the mobile application group (MAG) (n = 26) and the control group (CG) (n = 26). MAG received the two-month rehabilitation program via the developed application. CG was given paper-based exercise forms with the same protocol. Participants’ pain, expectation, and satisfaction levels were assessed with the Visual Analog Scale (VAS), and function was assessed with the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). All assessments were performed at baseline and after 8 weeks.ResultsBoth MAG and CG showed statistically significant improvement in VAS-rest, VAS-activity, WOMAC-pain, WOMAC-stiffness, WOMAC-function, and WOMAC-total scores (p < 0.05). However, there was no significant difference between MAG and CG for all pain and function scores (p > 0.05). In addition, no difference was observed between the two groups regarding expectation-satisfaction changes (p > 0.05). Besides, expectation-satisfaction change was not significantly different within each group (p > 0.05).ConclusionsRehabilitation presented with the mobile application was effective regarding pain and function. However, rehabilitation via mobile application did not provide additional contribution to pain, function, expectation-satisfaction compared to usual rehabilitation. Differences in change for function and satisfaction exceeded the minimal clinically important difference (MCID), favoring MAG.Trial registrationClinicalTrials.gov NCT06422910, 15 May 2024.Keywords: Function recovery, Knee osteoarthritis, Pain, Remote rehabilitation, Telerehabilitation Background Knee osteoarthritis (OA) is one of the most common chronic degenerative joint diseases. Knee OA is primarily seen in older adults [1]. Among individuals aged 60 years and older, 13% of women and 10% of men have been diagnosed with primary knee OA [2]. Although the social and economic costs of OA vary by society, country, culture, and gender, OA utilization of resources is significant worldwide [3]. In the long term, after treatment applications are performed in the clinic, patients perform exercises at home [4]. These exercises, which are given by physiotherapists, require active patient participation [5]. Numerous studies have shown that home exercise programs for knee OA lead to minimal to moderate improvements in pain, muscle strength, and physical function [6, 7]. OA is a disease that causes loss of proprioceptive sensation. A decrease in proprioceptive sensation can lead to weaker muscle function and increased disease severity, and progression. Decreases in muscle strength increase joint loading by causing misalignment and an increased Q angle. For these reasons, guidelines report the beneficial effects of conventional strengthening and stretching exercises on knee OA [8–10]. Physiotherapists develop home exercise programs for individuals with knee OA and monitor their rehabilitation remotely. Therefore, it is important that patients understand and can perform their exercises under optimal conditions [11]. Telerehabilitation applications offer many advantages during the treatment process. For an exercise program to be effective, it must have an appropriate progression. Exercise parameters should be organized according to changes in OA symptoms [12]. Telerehabilitation allows physiotherapists to easily communicate with their patients remotely and adjust their exercise programs. Maintaining communication between patients and physiotherapists is key to a successful rehabilitation process. Providing home exercise instructions verbally after a face-to-face session or through paper-based brochures can result in communication challenges between patients and therapists, such as patients forgetting or performing the exercises incorrectly or ineffectively. Furthermore, therapists cannot objectively verify that patients are carrying out these exercises appropriately at home because they cannot determine how much attention patients give to verbal or brochure-based instructions. Thus, therapists must rely solely on patients’ self-reported adherence [13–15]. Therefore, mobile applications that deliver detailed, video-based home exercise instructions may be an effective solution. Furthermore, these platforms allow therapists to monitor in real time whether patients have viewed the instructional videos, thereby enhancing the ability to objectively track adherence. According to the literature, patient education involving explanatory videos increases adherence to exercise, improves belief in recovery through an internal locus of control, and increases participation in rehabilitation [16]. Studies have shown that, in telerehabilitation, mobile applications are used more frequently than other impractical tools, such as web apps, social media, and links. Field studies in health and engineering sciences have revealed that mobile apps are more effective than web apps and manual methods [17, 18]. Therefore, our research focused on the effectiveness of mobile applications. Systematic reviews show that current mobile apps for knee and hip OA lack effective management plans, specific exercise prescriptions, and secure communication tools [19]. A similarly designed study provides exercise prescriptions with artificial intelligence [20]. However, it does not offer specific exercises based on a physiotherapist manual. Another study found that the application included nursing rehabilitation exercises, but did not provide video demonstrations [21]. In a recent intervention study, the effectiveness of the application was tested on only female individuals. While this study had an efficient design in terms of homogeneity, it lacked generalizability in terms of gender [22]. Due to the low sample size, the level of evidence for the outcomes in another pilot study could only be preliminary [23]. Two qualitative studies emphasize the importance of communication between clinicians and patients [24, 25]. Systematic reviews report that exercise programs administered via digital platforms at home for knee osteoarthritis provide modest improvements in pain, strength, and function, and that the effect size and durability are inconsistent across studies [26]. Although web-based iBEAT provides clinically meaningful benefits compared to usual care, it does not provide objective, therapist-verifiable measurements for monitoring progress [27]. Adherence to unsupervised home programs follows heterogeneous trajectories and is a major mediator of treatment effect, underscoring the need for platforms that both measure execution objectively and actively support sustained engagement. Our study fills this gap by developing a video-based exercise mobile application that standardizes exercise dosage and progression criteria, captures objective, therapist-verifiable application metrics using embedded video analytics, and links these metrics to clinician dashboards and predefined progression rules. Our research aimed to improve the design and methodological quality of the trial app by addressing the limitations of current mobile applications for knee OA. The key feature of our study is providing exercises to patients with knee OA through videos offering visual and auditory feedback. The app also facilitates communication with clinicians via an individual communication service. Our application stands out because it evaluates patients’ pain, functionality, expectations, and satisfaction presents the results to clinicians with statistical data. This study aims to develop a video exercise-based mobile application and assess its effectiveness in terms of pain, function, expectations, and satisfaction in patients with knee OA. Methods Mobile-app The developed video-exercise tracking application can be used with both Android and iOS operating systems. The developed application has three main functions, which are detailed below. Treatment Exercises related to knee OA were categorized according to their types (e.g., stretching, strengthening) and the body part. These video sections allow physiotherapists to filter the videos deemed appropriate. The exercise program can be added to any messages sent to the patients. Patients have the opportunity to access their own exercises remotely with visual, auditory, and written commands. Assessment The Turkish versions of the most widely used evaluation questionnaires developed for knee OA and adapted to various languages, whose validity and reliability have been demonstrated, are sent to the patients by the clinician via mobile application at appropriate times. The progress of the treatment is monitored. The physiotherapist organizes the exercise program following the evaluation results. All the data related to the evaluation can be automatically processed into the database. Treatment outcome-communication Individual exercise programs can be sent and edited through the messaging system, enabling communication between the patient and the clinician before, after, and throughout the treatment program. Evaluation and patient information statistics can also be carried out through this messaging system. In addition, within the scope of the automatic reminder feature of the mobile application, warnings to be sent to the patient before exercise can also be carried out through this communication network. The diagram of the interfaces of the mobile application is given in Fig. 1 (see Fig. 1). The sample screenshots from the developed mobile application Application development The application has components called front-end, back-end, and database. The application was developed using a cross-platform React Native framework and JavaScript libraries. This choice was preferred because of the mobile application’s more efficient performance and speed [28]. Figure 2 shows the flowchart of the mobile application developed (see Fig. 2). The mobile application was subjected to the first test with alpha version tests. At this stage, the database, back-end, and front-end functionalities were checked, and it was tested whether the visible part of the mobile application, the code part running in the back, and the database part where the information is stored were working correctly. With the security and access control test, any possible error and vulnerability problems during the operation of the mobile application were checked. A randomized controlled study was conducted within the scope of beta version tests. Mobile application flowchart Randomized controlled trial design and participants The randomized controlled trial was carried out with 52 individuals with knee OA who were followed up in Fethiye State Hospital Orthopaedics and Traumatology Outpatient Clinic. The study was carried out in accordance with ethical principles and the Helsinki Declaration. The patient’s written consent was obtained. The study protocol was approved by the İzmir Bakırçay University Non-Invasive Clinical Research Ethics Committee (No: 968). The protocol was recorded on clinicaltrials.gov (ClinicalTrials.gov Identifier: NCT06422910/Record Verification Date: 15 May 2024). The present study follows the “CONsolidated Standards of Reporting Trials (CONSORT)” guidelines. Inclusion criteria were (1) being diagnosed with primary knee OA according to the American College of Rheumatology (ACR) criteria, (2) having the ability to use the web application and agreeing to use the application. Exclusion criteria were (1) individuals with poor cognitive function or inadequate reading skills, (2) individuals with neurological disorders that may affect mobility skills, and (3) individuals with knee joint surgery and traumatic injuries. Sample size The sample size calculation utilized a Cohen’s d of 0.80, determined by considering the WOMAC score from a previous similar study [29]. The effect size = 0.80, an error probability of 0.05, and a statistical power of 0.80 were calculated for the sample size using the Gpower software. The result showed that at least 21 cases would be sufficient. Therefore, our aim was to include a total of at least 42 individuals. According to the post-hoc power analysis, the observed between-group mean difference in the total WOMAC change was − 1.7 (95% CI − 7.77 to 4.37). The corresponding Cohen’s d was − 0.16. With n = 26 participants per group, the post-hoc power to detect the observed effect size (d = 0.16) was approximately 0.08. These calculations indicate that the sample size of the present study has limited power to detect small effects between groups. Randomization and blinding Participants were randomly allocated into two groups: the mobile application group (MAG) (n = 26) and the control group (CG) (n = 26) (see Fig. 3). The National Institutes of Health National Cancer Institute Clinical Trial Randomisation Tool was used for randomization [30]. The assignments created by this software using the asymptotic maximal procedure method determined the individuals to be included in the intervention and control groups [31]. There was no blinding in present study. Patient recruitment process of the research Interventions The developed mobile application was used to follow up with individuals in MAG. The application sent Exercise videos to the individuals, and remote consultation was provided. Individuals in the conventional rehabilitation group were given paper-based exercise forms with pictures and explanations. Control group In the CG, exercises were taught to the patient in the clinic after the initial evaluation. An information sheet (prepared by the researchers) schematizing and explaining the exercises was given. The exercise program included terminal knee extension, straight leg raises, knee flexion and extension in a sitting position, mini squats, and stretching exercises [32]. The exercises were planned to last approximately 15–20 min. The exercises were prescribed as 8–10 repetitions and two sets. Individuals were asked to perform the exercises five days a week for eight weeks. All participants were given the same number, intensity, and dosage of seated knee extension, isometric quadriceps femoris activation with a roll under the knee (knee pushing), straight leg raises, mini squats against the wall, and self-administered hamstring stretches. Exercise tracking was provided by contacting the patient by phone. Mobil application group In the mobile application exercise group, exercises were delivered via video mobile application [32]. The exercise videos included videos prepared for the web application, which was developed in the researchers’ previous study. The exercise program content, duration, and number of repetition sets were the same as the CG. Individuals were asked to perform the exercises five days a week for eight weeks. All participants were given the same number, intensity, and dosage of seated knee extension, isometric quadriceps femoris activation with a roll under the knee (knee pushing), straight leg raises, mini squats against the wall, and self-administered hamstring stretches. Individuals in the mobile application group actively used the reminder and messaging features of the application. Exercise tracking was carried out via a mobile application. Data collection tools Socio-demographic information, clinical features of OA, medical history, family history, and habits of the individuals were recorded. Pain, functional status, expectations, and satisfaction assessments were performed at the beginning of the study and the end of the 8-week intervention. Details of the assessment tools are presented below. Visual analog scale The patient is asked to mark the pain felt on a 0–10 cm straight line or numeric scale (0: no pain, 10: unbearable pain). Our study used a numeric Visual Analog Scale (VAS). VAS was assessed at baseline and after 8 weeks. VAS scores of 3.4 or lower indicate mild pain, scores between 3.5 and 7.4 represent moderate pain, and scores of 7.5 or higher reflect severe pain [33, 34]. The digital version of the VAS has been reported to be reliable, valid, and sensitive for measuring pain intensity. The test-retest reliability of the VAS, measured by intraclass correlation coefficients, has been reported as r = 0.96. Concurrent validity ranges from r = 0.84 to 0.97. The responsiveness coefficient indicated small effect size (0.20–0.50) [35]. Western Ontario and McMaster universities osteoarthritis index The functionality level in the patient group with knee OA was measured using Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC). WOMAC was assessed at baseline and after 8 weeks. WOMAC consists of three main headings: pain intensity, stiffness, and physical function. The individuals included in the study were asked to score each item between 0 (no pain, stiffness, difficulty) and 5 (very severe pain, stiffness, difficulty). The total score varies between 0 (no disability) and 96 (complete disability) [36]. The WOMAC Cronbach’s alpha coefficients for reliability were as follows: Pain: 0.75, 0.81; Stiffness: 0.71, 0.76; Physical function: 0.94, 0.96. In terms of validity, the frequency distribution of the scores on each subscale of the WOMAC index was quite symmetrical and did not show any significant skewness (range between + 0.13 and − 0.01). A fairly good correlation has been reported between the Short Form-36 and WOMAC. A good correlation was found between the WOMAC total score and the Lequesne index, and it was reported that the pain and physical function subscales of the WOMAC index were the most responsive subscales [36]. Expectation and satisfaction assessment Expectation and satisfaction assessments were conducted by VAS. On a 0–10 cm straight line or numeric scale, the patient was asked to tick the level of expectation before treatment and satisfaction after treatment. Expectation levels were evaluated during the initial assessment, whereas satisfaction levels—reflecting the degree to which these expectations were fulfilled—were assessed during the follow-up evaluation (after 8 weeks). A score of 0 represents the lowest expectation and satisfaction, and 10 shows the highest expectation or satisfaction [33]. Statistical analysis SPSS v25 for Mac software was utilized for data analysis. The One-Sample Kolmogorov-Smirnov test was performed for homogeneity testing. As a result of this test, non-parametric tests were employed for all analyses. In comparing baseline characteristics between MAG and CG, the Mann-Whitney U test was used for continuous variables, while the Pearson chi-square test was used for categorical variables. The two-group comparison was analyzed using the Mann-Whitney U test. The formula was used to calculate the effect size in the Mann-Whitney U and Wilcoxon Signed Rank tests. The Wilcoxon Signed Rank Test was used to measure MAG and CG within groups. Minimal clinically important differences (MCIDs) were calculated by multiplying the effect size and the standard deviation of the first measurement for within-group comparisons. For between-group comparisons, they were calculated by multiplying the effect size and the pooled standard deviations. Correlation analyses were performed using Spearman’s correlation. The margin of error was accepted as 0.05 in all statistics. Results The mean ages of the individuals included in the study were 57.2 ± 8.1 and 61.7 ± 8.6 years for MAG and CG, respectively. The groups were not statistically different in terms of baseline characteristics. The two groups were homogeneous regarding individual parameters, except for the affected limb (p = 0.041). The mean Kellgren-Lawrence score representing the severity of OA was 2.3 ± 1.0 and 2.8 ± 0.7 for MAG and CG, respectively. Data on the individual characteristics of the groups are presented in Table 1. Table 1Patient characteristics in MAG and CGMAG (n = 26)CG (n = 26) p Gender (women/men, n)19/719/71.000aAge (years, mean ± SD)57.2 ± 8.161.7 ± 8.60.060bBody mass index (kg/m2, mean ± SD)30.6 ± 5.230.8 ± 4.90.942bAffected extremity (right/left/bilateral, n)4/4/187/10/90.041aDominant extremity (right/left, %)26/024/20.149aEmployment (yes/no, n)10/168/180.560aEducation (high school and below/high school above, n)22/424/20.385aSmoking (yes/no, %)5/214/220.714aComorbid disease (yes/no, %)13/1314/120.781aDuration of symptom (months, mean ± SD)28.9 ± 70.218.2 ± 20.30.846bKellgren-Lawrence Score (mean ± SD)2.3 ± 1.02.8 ± 0.70.061bn number, SD standard deviation, MAG mobile application group, CG control group. a: Pearson’s chi-square test, b: Mann-Whitney U test Patient characteristics in MAG and CG n number, SD standard deviation, MAG mobile application group, CG control group. a: Pearson’s chi-square test, b: Mann-Whitney U test Table 2 presents the groups’ results regarding clinical measurements and expectation-satisfaction. For the VAS-rest score, both MAG (p = 0.002) and CG (p = 0.002) showed statistically significant improvement after eight weeks of intervention. However, there was no significant difference between MAG and CG for rest pain (p = 0.832). The same outcomes were confirmed for the VAS-activity score. While no significant difference was observed between MAG and CG regarding VAS-activity score (p = 0.932), within-group changes were significant in both groups (pmag=0.0003, pcg=0.001). There was no statistically significant difference between MAG and CG regarding WOMAC and its sub-scores (p > 0.05). No difference was observed between the two groups regarding expectation-satisfaction changes in the patients (p > 0.05). In intergroup comparisons, all effect sizes were at the small effect level. The intra-group change values of WOMAC-pain, WOMAC-stiffness, WOMAC-function, and WOMAC-total scores improved significantly in both groups (p < 0.05). Besides, expectation-satisfaction change was not significantly different within each group (p > 0.05). According to the intra-group analyses, all results except for expectation and satisfaction levels were at a medium effect size level; only expectation and satisfaction levels were at a small effect size level. Table 2The comparison of the groups in terms of clinical outcomes and expectation/satisfactionMAG (n = 26)CG (n = 26)BCa 95% CI(lower)BCa 95% CI(upper)p (between)(Cohen’s r)MCID(within-group)MCID(between-group)VAS-rest1 st assessment2.8 ± 2.13.6 ± 2.6−210.402a(−0.116)MAG: 0.89CG: 1.130.272nd assessment1.8 ± 2.02.5 ± 2.2−210.247a(−0.161)∆ (change)−1.0 ± 1.4−1.1 ± 1.5−0.0000099710.832a(−0.029)p (within)(Cohen’s r) 0.002 b (−0.422) 0.002 b (−0.435) VAS-activity1 st assessment7.0 ± 1.77.7 ± 1.7−20.000017840.136a(−0.207)MAG: 0.83CG: 0.810.352nd assessment5.2 ± 2.45.8 ± 2.5−210.507a(−0.092)∆ (change)−1.7 ± 1.7−1.9 ± 2.1−110.932a(−0.012)p (within)(Cohen’s r) 0.0003 b (−0.491) 0,001b (−0.475) WOMAC-pain1 st assessment32.5 ± 12.537.8 ± 13.7−150.000059970.188a(−0.183)MAG: 5.9CG: 7.112.42nd assessment21.1 ± 14.626.1 ± 16.8−1050.240a(−0.163)∆ (change)−11.3 ± 13.2−11.7 ± 12.0−550.772a(−0.04)p (within)(Cohen’s r) 0.001 b (−0.472) 0.0001 b (−0.519) WOMAC-stiffness1 st assessment21.4 ± 14.625.0 ± 15.8−12.50.000032160.438a(−0.108)MAG: 5.97CG: 7.21.642nd assessment13.9 ± 15.116.3 ± 15.3−12.50.000051460.510a(−0.091)∆ (change)−7.5 ± 10.3−8.6 ± 9.8−0.0000414750.698a(−0.054)p (within)(Cohen’s r) 0.003 b (−0.409) 0.001 b (−0.456) WOMAC-function1 st assessment39.1 ± 10.543.0 ± 15.3−11.830.355a(−0.128)MAG: 5.81CG: 7.791.682nd assessment27.1 ± 15.333.4 ± 18.1−14.730.220a(−0.17)∆ (change)−12.0 ± 11.1−9.5 ± 11.2−7.330.345a(−0.131)p (within)(Cohen’s r) 0.00006 b (−0.553) 0.0002 b (−0.509) WOMAC-total1 st assessment36.2 ± 10.840.3 ± 14.4−12.53.10.355a(−0.128)MAG: 5.98CG: 7.821.632nd assessment24.7 ± 14.730.4 ± 17.1−13.53.10.230a(−0.166)∆ (change)−11.5 ± 10.9−9.8 ± 10.9−6.54.10.595a(−0.074)p (within)(Cohen’s r) 0.00006 b (−0.554) 0.00009 b (−0.543) VAS-expectation/satisfaction1 st assessment(expectation)5.1 ± 1.34.4 ± 1.3−0.000022320.055a(−0.266)MAG: 0.23CG: 0.070.352nd assessment(satisfaction)5.8 ± 2.94.2 ± 2.6−0.0000480630.060a(−0.261)∆ (change)0.6 ± 2.4−0.1 ± 2.4−120.272a(−0.152)p (within)(Cohen’s r)0.202b(−0.177)0,694b(−0.055)n number, SD standard deviation, MAG mobile application group, CG control group, VAS Visual Analog Scale, WOMAC Western Ontario and McMaster Universities Osteoarthritis Index, BCa 95% CI Bias-Corrected and Accelerated 95% Confidence Interval, MCID Minimal Clinically Important Difference. a: Mann-Whitney U test, b: Wilcoxon Signed Rank test The comparison of the groups in terms of clinical outcomes and expectation/satisfaction 0.402a (−0.116) MAG: 0.89 CG: 1.13 0.247a (−0.161) 0.832a (−0.029) p (within) (Cohen’s r) 0.002 b (−0.422) 0.002 b (−0.435) 0.136a (−0.207) MAG: 0.83 CG: 0.81 0.507a (−0.092) 0.932a (−0.012) p (within) (Cohen’s r) 0.0003 b (−0.491) 0,001b (−0.475) 0.188a (−0.183) MAG: 5.9 CG: 7.11 0.240a (−0.163) 0.772a (−0.04) p (within) (Cohen’s r) 0.001 b (−0.472) 0.0001 b (−0.519) 0.438a (−0.108) MAG: 5.97 CG: 7.2 0.510a (−0.091) 0.698a (−0.054) p (within) (Cohen’s r) 0.003 b (−0.409) 0.001 b (−0.456) 0.355a (−0.128) MAG: 5.81 CG: 7.79 0.220a (−0.17) 0.345a (−0.131) p (within) (Cohen’s r) 0.00006 b (−0.553) 0.0002 b (−0.509) 0.355a (−0.128) MAG: 5.98 CG: 7.82 0.230a (−0.166) 0.595a (−0.074) p (within) (Cohen’s r) 0.00006 b (−0.554) 0.00009 b (−0.543) 1 st assessment (expectation) 0.055a (−0.266) MAG: 0.23 CG: 0.07 2nd assessment (satisfaction) 0.060a (−0.261) 0.272a (−0.152) p (within) (Cohen’s r) 0.202b (−0.177) 0,694b (−0.055) n number, SD standard deviation, MAG mobile application group, CG control group, VAS Visual Analog Scale, WOMAC Western Ontario and McMaster Universities Osteoarthritis Index, BCa 95% CI Bias-Corrected and Accelerated 95% Confidence Interval, MCID Minimal Clinically Important Difference. a: Mann-Whitney U test, b: Wilcoxon Signed Rank test Tables 3 and 4 show the correlation coefficients between expectation and satisfaction scores and pain and function values of patients in MAG and CG. There was a low correlation between the expectations of the individuals in MAG and all pain and function clinical scores (r < 0.35, p > 0.05). However, there was a moderate correlation between the satisfaction score of the individuals in MAG and the WOMAC-stiffness change value (r=−0.466, p = 0.016). In addition, there was a strong correlation between the satisfaction values of MAG and VAS-rest, VAS-activity, WOMAC-pain, WOMAC-function, and WOMAC-total change scores (r > 0.5, p < 0.001). Likewise, there was a low correlation between the expectation scores of individuals in CG and clinical measurements (r < 0.35, p > 0.05). However, there was a strong correlation between the satisfaction score of CG and all clinical parameters (r > 0.5, p < 0.001). Table 3The relationship between expectation/satisfaction with pain and the function of the MAGn: 26VAS-expectationVAS-satisfactionVAS-rest (change value)−0.138−0.585**VAS-activity (change value)−0.165−0.832**WOMAC-pain (change value)−0.136−0.697**WOMAC-stiffness (change value)0.085−0.466*WOMAC-function (change value)−0.195−0.731**WOMAC-total (change value)−0.203−0.743**n number of patients, *: p < 0.05, **: p < 0.001All analyses were performed using Spearman’s correlation analysis The relationship between expectation/satisfaction with pain and the function of the MAG n number of patients, *: p < 0.05, **: p < 0.001 All analyses were performed using Spearman’s correlation analysis Table 4The relationship between expectation/satisfaction with pain and the function of the CGn: 26VAS-expectationVAS-satisfactionVAS-rest (change value)0.091−0.774**VAS-activity (change value)−0.002−0.843**WOMAC-pain (change value)0.113−0.706**WOMAC-stiffness (change value)0.117−0.732**WOMAC-function (change value)−0.060−0.811**WOMAC-total (change value)0.020−0.806**n number of patients, *: p < 0.05, **: p < 0.001All analyses were performed using Spearman’s correlation analysis The relationship between expectation/satisfaction with pain and the function of the CG n number of patients, *: p < 0.05, **: p < 0.001 All analyses were performed using Spearman’s correlation analysis Additionally, although there were no differences between groups at baseline, we performed an ANCOVA analysis adjusted for baseline values. ANCOVA adjusting for baseline VAS-rest values showed no significant group effect on follow-up VAS-rest scores, F(1, 53) = 0.028, p = 0.867, partial eta2 = 0.001. Adjusted means were MAG = 2.30 (95% CI 1.79–2.80) and CG = 2.24 (95% CI 1.71–2.76), indicating no clinically meaningful difference between groups after controlling for baseline imbalance. Pairwise comparisons of adjusted means indicated no significant between-group difference (mean difference = 0.062, p = 0.867, 95% CI − 0.672 to 0.795). After controlling for baseline VAS-activity scores, no significant group effect was observed on follow-up VAS-activity, F(1, 53) = 0.252, p = 0.618, partial eta² = 0.005. Adjusted means were MAG = 5.75 (95% CI 5.02–6.49) and CG = 5.49 (95% CI 4.73–6.25), indicating that baseline imbalance did not materially influence the between-group comparison. Pairwise adjusted mean difference was small and nonsignificant (mean difference = 0.266, 95% CI − 0.797 to 1.329, p = 0.618), consistent with the overall ANCOVA result. Similarly, there were no significant differences in the adjusted ANCOVA analysis for WOMAC-pain, WOMAC-stiffness, WOMAC-function, WOMAC-total, and expectation-satisfaction scores (p1 = 0.858, p2 = 0.774, p3 = 0.57, p4 = 0.671, p5 = 0.314). Finally, the VAS-rest and expectation-satisfaction changes according to MCID values were very similar to the MCID in CG. The change values were ∆= −1.1 ± 1.5 and − 0.1 ± 2.4, respectively. The MCID values were 1.13 and 0.07, respectively. In contrast, the change values for all other parameters were greater than the MCID value identified within the group. The differences in change between groups for WOMAC-function, WOMAC-total, and expectation-satisfaction were ∆∆= 2.5, 1.7, and 0.7, respectively, in favor of the MAG group. The MCID values were 1.68, 1.63, and 0.35, respectively. Therefore, the differences in change between groups in WOMAC-function, WOMAC-total, and expectation-satisfaction were greater than the MCID. The differences in change between groups for all other parameters were less than the MCID values. Discussion This study aimed to develop a video exercise-based exercise-tracking mobile application. In the clinical purpose of the study, the effect of the developed mobile application on pain, function, expectation, and satisfaction in individuals with knee OA was investigated. According to the study results, rehabilitation presented with the mobile application was effective regarding pain and function. Both the rehabilitation group offered via the mobile app and the control group showed improvement with a moderate effect size. However, rehabilitation via mobile application did not provide additional contribution to pain and function compared to usual rehabilitation. Additionally, the mobile app-based rehabilitation group did not provide any additional benefit in terms of expectation and satisfaction compared to the control group. Since the mobile application group was not superior to the control group, the effect sizes were small. Since long-term follow-up of at least eight weeks is required for exercise programs to be effective in individuals with knee OA [37], patients are usually followed up by clinicians within the scope of home programs. In this context, home exercise prescriptions are provided to patients within the scope of usual rehabilitation ordered on paper [7]. Physiotherapists have difficulty following their patients remotely. It is also known that patients have difficulties in learning and applying exercises correctly [38]. In recent years, thanks to the web-based telerehabilitation technique, patients’ communication with the clinician can be provided more effectively. In addition, it has been reported that video-based applications give more effective results owing to multimedia content [14–16]. However, the current literature has not sufficiently investigated the effectiveness of mobile application-based telerehabilitation. We assumed that the mobile application we developed could obtain more efficient results than web applications. Studies in engineering sciences and health services have emphasized some advantages of mobile applications over web-based software [17, 18]. A cross-sectional study reported that patients wanted to see mobile applications more than web applications [18]. In addition, qualitative studies have emphasized mobile applications’ importance in patient-clinician communication [24, 25]. In this respect, the originality of our study regarding the intervention technique should be emphasized. A recent systematic review states that mobile applications in individuals with knee and hip OA do not provide specific exercise plans and do not provide safe communication tools [19]. Based on this issue, we aimed to strengthen clinician-patient communication with the messaging module of the mobile application we developed within the scope of our study. Additionally, we aimed to improve exercise tracking and enhance patients’ internal compliance with home exercises by maintaining communication with them through a mobile application. Despite these precautions, we could not obtain any additional advantage in individuals’ pain and function scores compared to standard rehabilitation applications. Moreover, the expectation and satisfaction parameters of the individuals in the two groups were similar. These results suggest that rehabilitation offered by mobile applications may not provide a clinically advantageous health service, although it offers specific exercise prescriptions and communication with the clinician. Nevertheless, considering the advantages of telerehabilitation in terms of cost-effectiveness in previous studies and the transport advantage it provides to individuals with disabilities [39, 40], the study’s results may provide practical outputs to clinical practice in terms of the use of mobile application-based telerehabilitation in different clinical conditions. In a randomized controlled study with a similar design, an artificial intelligence-based mobile application was developed to improve exercise accuracy and quality of life in individuals with knee OA. After a four-week exercise program, individuals were evaluated regarding range of motion, pain symptoms, physical activity, and quality of life [20]. The results of this study reported the effectiveness of the AI-based mobile application in terms of exercise accuracy and outcomes. However, it is also seen that the mobile application does not provide additional advantages in some clinical parameters related to range of motion, quality of life, and function. Considering these clinical outcomes comparable to the results of our study, it can be inferred that mobile applications containing manual methods or artificial intelligence-based technologies offer an alternative health service to classical rehabilitation. In previous study conducted with eighty participants, the effectiveness of an application containing a phone call messaging patient recording system was evaluated. In this intervention study method, the exercise content of the application was not mentioned. Nevertheless, the results reported the mobile application’s effectiveness in improving individuals’ participation in rehabilitation and clinical parameters [21]. However, it is impossible to discuss the effectiveness of a video exercise protocol as in our study. Another randomized controlled study focused on the effect of mobile applications on physical function only in individuals with female OA. Sixty-four female individuals were included in the study, and the training program offered within the scope of routine care was given with a mobile application [22]. However, this study did not present details about the video exercise protocol. Although the advantage of the mobile application over average rehabilitation in terms of function and quality of life is emphasized in this study, it should be said that there is no design comparable to the results of our study. A pilot intervention study focused on the level of exercise participation of the mobile application in individuals with knee OA. After the 12-week intervention program, it was emphasized that the participation of individuals was over 80%. This study emphasized the effectiveness of pain, function, and satisfaction, and the usability of the mobile application was underlined, especially in providing home programs. Although mobile applications did not provide additional advantages in our study, they effectively provided home program prescriptions regarding pain function [23]. However, our study could not evaluate individuals’ participation in rehabilitation. Future studies may provide analysis results based on exercise participation. The MAG group demonstrated functional and satisfaction improvements that exceeded the MCID than CG, reflecting clinically meaningful benefits. Similarly, pain and stiffness scores also showed a trend toward improvement in the MAG group but not in the CG. These findings indicate that the most pronounced improvement effects were observed in functional capacity and satisfaction. Overall, the MCID data indicates that the MAG intervention provided patients with concrete benefits in terms of function and satisfaction. Clinical implications Some clinical outcomes were identical for MAG and CG in this trial. Our clinical experience in telerehabilitation shows that the barriers to using mobile application-based telerehabilitation are still a problem. Patients may not sufficiently provide the expected exercise adherence-adaptation. In addition, due to the nature of our sample group, the average age was relatively high, which is inversely proportional to technology literacy. The difficulty of using the mobile application in our patients may have led them to seek help from their relatives and, even worse, may have limited the use of the application. The outcome measures of the current study were based on patient-reported questionnaires. The qualitative nature of these outcome measures may have affected the objectivity of our results as many conditions influence them and may have contributed to the similar results between MAG and CG. The mean Kellgren-Lawrence stages were relatively high in our study, which may have limited the benefits that could have been achieved with physiotherapy and rehabilitation. Essentially, we can conclude that mobile application home exercise monitoring can replace conventional methods in this population. To achieve more effective results, it will be crucial to use applications with simple interfaces and target a sample size with high technology literacy. However, the necessity of closely monitoring usability and exercise adherence must be balanced. By doing so, we can ensure that the potential benefits of telerehabilitation are fully realized, leading to improved health outcomes, reduced costs, and increased practicality. This underscores the importance of these actions in the context of telerehabilitation. Limitations Although our study has a randomized controlled design, it does not include single or double-blind evaluation, treatment, or sampling. This issue indicates that there may be biased results in international guidelines [41]. However, we believe we provided a standardized clinical evaluation in our study with a single physiotherapist undertaking the evaluation procedures. Secondly, we could not associate how often patients watched exercise videos and communicated with the clinician through our mobile application with our analyses. Since participants are recruited from a single center, this may limit the external validity. Participants were not prevented from receiving other treatments; this indicates that participants may have received other treatments during our exercise protocol. A limitation of this study is that the observed between-group effect for the WOMAC-total was smaller than the effect size (d = 0.80) used in the pre-calculated power calculation. The post-hoc power obtained for the observed effect (d = 0.16) was only 8%, which limited the ability to detect statistically significant between-group differences. Future studies should employ larger sample sizes with more conservative effect sizes. Future studies may provide predictive inferences based on statistics such as frequency of use when evaluating the efficiency of telerehabilitation. Finally, it may be possible to provide the effectiveness of the mobile application, which is the primary purpose of our study, more effectively by comparing it with the web application. In this respect, future studies can be designed with a control mechanism comparing mobile versus web applications. Conclusions This study aimed to develop a video exercise-based exercise-tracking mobile application. In the clinical purpose of the study, the effect of the developed mobile application on pain, function, expectation, and satisfaction in individuals with knee OA was investigated. According to the study results, rehabilitation presented with the mobile application was effective regarding pain and function. However, rehabilitation via mobile application did not provide additional contribution to pain and function compared to usual rehabilitation. Additionally, the mobile app-based rehabilitation group did not provide any additional benefit in terms of expectation and satisfaction compared to the control group. The adjusted analysis using baseline scores as a covariate showed that the groups were not superior to each other, confirming that the baseline imbalance did not affect the post-intervention outcomes. According to MCID, although CG did not show a clinically important difference in rest pain and satisfaction levels in within-group changes, MAG showed a clinically important difference. The differences in change between groups for function and satisfaction exceeded the MCID, favoring the MAG group. Acknowledgements The authors thank TUBITAK for their supports. Abbreviations American college of rheumatology Control group Osteoarthritis Mobile application group Minimal clinically important difference Visual analog scale Western Ontario and McMaster universities osteoarthritis index Authors’ contributions F.Ö.: Conceptualization, Funding acquisition, Methodology, Writing – original draft, Writing – review & editing, Data curation, Formal analysisB.U.T.: Conceptualization, Methodology, Project administration, Writing – review & editing, SupervisionO.Ç.: Methodology, Writing – review & editingB.E.S.: Methodology, Software, Writing – original draftN.T.: Methodology, Visualization, Writing – review & editingZ.S.: Methodology, Supervision, Writing – review & editingİ.Ç.: Writing – original draftG.C.: Writing – original draftİ.T.: Investigation, Writing – original draft. Funding This study was supported by Scientific and Technological Research Council of Turkey (TUBITAK) under the Grant Number 123S170. Data availability All data generated or analyzed during this study are included in this published article. Declarations Ethics approval and consent to participate The study was carried out in accordance with ethical principles and the Helsinki Declaration. The study protocol was approved by the İzmir Bakırçay University Non-Invasive Clinical Research Ethics Committee (No: 968). Informed consent of the patients was obtained. The protocol was recorded on clinicaltrials.gov (ClinicalTrials.gov Identifier: NCT06422910/Record Verification Date: 15 May 2024). Consent for publication No individual or identifiable data is being published as part of this manuscript. All participants have provided written informed consent for their personal or clinical information to be published in this study. 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 All data generated or analyzed during this study are included in this published article.