This study adopted a clinician-centered, cross-sectional evaluation design to assess exercise prescription apps intended for professional use. A multidimensional evaluation framework was applied to reflect the key components of clinical exercise prescription practice (Figure 1).

The framework included four domains: (1) clinical integrity, (2) intervention fidelity, (3) behavioral mechanism, and (4) digital usability. These domains were operationalized using established instruments: the FITT and FITT-VP principles, CERT, BCTTv1, and MARS.

This study did not involve human participants, human data, or human biological materials. Therefore, institutional ethics board review and approval were not required.

This review adhered to the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) protocol, as detailed in the PRISMA 2020 statement (Checklist 1) [31]. A systematic search of Google Play and the Apple App Store was conducted on July 13, 2024, using Taiwanese regional settings. The following search terms were used: “exercise prescription AND doctor,” “exercise prescription AND therapist,” and “exercise prescription AND rehabilitation.” Duplicate apps were removed.

Apps were eligible if they were free, available in English or Chinese, designed for adult users, and not disease specific; did not require additional equipment; and allowed clinicians to prescribe exercise programs. To ensure market relevance, only apps available on both platforms with more than 10,000 downloads were included. A comprehensive and transparent description of this search process is provided in Multimedia Appendix 1.

The initial search yielded a total of 1020 apps (each search term yielded 240 apps on Google Play and 100 on the App Store). After removing duplicates and applying the inclusion and exclusion criteria, 6 exercise prescription apps were included in the final analysis (Figure 2). All included apps were available on both platforms; were free to download; supported clinical exercise prescription; and had exceeded 10,000 downloads at the time of assessments.

As of July 29, 2024, the 6 apps included in our study demonstrated a wide range of installations, with numbers ranging from 10,000 to 1 million downloads (Table 1). The app developers were based in the United Kingdom, United States, Canada, and Australia. All apps were initially released between 2017 and 2020 and had all been updated in 2023 or 2024.

All 6 apps met the FITT criteria, as they explicitly specified the exercise frequency, intensity, time, and type within prescribed exercise programs (Table 2). In contrast, none of the apps met the FITT-VP criteria. Across all apps, explicit guidance on progression of exercise volume or intensity over time was absent, and no app provided decision rules for adjusting exercise prescriptions based on user performance or program duration.

Intervention fidelity varied across CERT domains (Table 3 and Figure 3). All 6 apps (100%) achieved complete reporting for core structural elements, including materials, provider qualifications, exercise setting, and dosage. Within the delivery domain, all apps specified whether exercises were performed individually or in groups and whether sessions were supervised or unsupervised. Of the 6 apps, 5 apps (83%) reported methods for assessing exercise adherence, and 4 apps (67%) included nonexercise components such as educational or dietary content.

In contrast, no app (0%) reported motivational strategies, decision rules for exercise progression, or descriptions of how exercise programs were progressed over time. Similarly, none of the apps described whether or how exercises were tailored to individual users or how starting exercise levels were determined. Reporting of adverse events was absent across all apps, and none documented whether the prescribed intervention was delivered as planned. The comprehensive table with respective scoring can be found in Multimedia Appendix 3. Interrater reliability for CERT assessment was high (Cohen κ=0.88).

Assessment of overall app quality using MARS yielded a mean score of 3.66 (SD 0.25) across the 6 apps, with individual app scores ranging from 3.36 to 4.00 (Table 6 and Figure 4). Among MARS domains, engagement received the lowest ratings, with a mean score of 2.50 (SD 0.28; range 2.10‐2.80), indicating limited use of interactive, motivational, or personalized features. In contrast, functionality was rated highest, achieving a mean score of 4.36 (SD 0.22; range 4.00‐4.63), reflecting stable performance, ease of navigation, and technical reliability across all apps. Aesthetic quality was rated moderately to highly, with a mean score of 3.89 (SD 0.48; range 3.17‐4.50), and information quality was rated similarly, with a mean score of 3.90 (SD 0.18; range 3.58‐4.07). Comprehensive tables with both reviewer and respective scoring results are provided in Multimedia Appendices 5 and 6. The interrater reliability of MARS was assessed using the ICC (2-way random effects, absolute agreement, average measures). The resulting ICC was 0.71 (95% CI 0.59‐0.81), indicating good reliability.

Our review of the evidence item revealed that by July 2024, only 2 apps among the cohort had undergone scientific evaluation, suggesting a lack of empirically validated health apps in the current market.

This study uses an integrated, clinician-centered, multidimensional evaluation framework to provide a novel professional perspective on how exercise prescription apps operate in real-world clinical contexts. Overall, popular, no-cost exercise prescription apps were found to meet the basic structural requirements for exercise prescription but fell short of key standards required for independent clinical use. Although prescriptions consistently specified core FITT elements, the absence of explicit progression and individualized adjustment limited alignment with established clinical practice. While multiple BCTs were identified, those critical for sustaining exercise behavior in unsupervised or remote settings were frequently lacking. Despite strong technical functionality and high information quality, low engagement scores suggest that usability alone is insufficient to support long-term adherence.

When considered collectively across domains, these apps nonetheless demonstrate operational strengths, including structured translation of prescriptions into actionable tasks, continuous visibility of adherence, low cognitive load instruction delivery, and delegation of routine monitoring. Taken together, these findings highlight a gap between clinical expectations for exercise prescription and the current capabilities of widely used digital tools, supporting the role of exercise prescription apps as adjuncts within clinician-guided care rather than as stand-alone prescribing solutions.

This study introduces a clinician-centered evaluation framework that integrates clinical exercise prescription standards, behavior change theory, and digital usability into a unified, practice-oriented model. Unlike prior evaluations of physical activity and mHealth apps that have primarily examined usability, engagement, or behavior change features in isolation [26-28], this framework situates app assessment within the logic of real-world clinical exercise prescription and decision-making. By aligning established instruments with sequential stages of professional practice, including prescription structure and progression logic (FITT and FITT-VP), transparency and reproducibility of intervention delivery (CERT), support for exercise initiation and maintenance (BCTTv1), and interface quality and engagement (MARS), the framework enables clinicians to directly compare app features with clinical expectations. In doing so, it provides a shared reference to guide both informed clinical selection and future app development.

Unlike general physical activity apps that often fail to meet the basic FITT criteria [24,42], the specialized apps in our study successfully incorporated these fundamental elements. However, the primary clinical limitation lies in their failure to satisfy the full FITT-VP framework, specifically regarding structured progression and individualized adjustment. In established exercise prescription practice, progression is essential for achieving physiological adaptation while minimizing injury risk, particularly by avoiding abrupt increases in exercise volume or intensity [43,44]. Clear guidance on how and when to advance exercises is also critical for informed consent and may facilitate patient engagement and adherence.

In addition, limited support for tailoring exercise programs to individual needs constrains the safe use of these tools in populations such as older adults and individuals with chronic diseases, for whom inappropriate exercise loading may increase the risk of adverse events [45]. The absence of explicit information regarding potential exercise-related risks further complicates informed decision-making by both clinicians and patients. To optimize the integration of exercise prescription apps into clinical practice, more comprehensive and transparent intervention descriptions are required to support individualized care and the effective translation of evidence-based exercise prescription into real-world settings.

The evaluation of intervention fidelity via the CERT domains revealed a stark disparity between the reporting of static structural elements and dynamic clinical processes. While all apps achieved 100% compliance in describing core components such as materials, provider qualifications, and settings, there was a universal failure (0%) to report motivational strategies, progression logic, or tailoring mechanisms. This deficiency mirrors findings from previous systematic reviews in pregnancy [46] and musculoskeletal rehabilitation [47], which similarly identified a pervasive lack of customization features in mHealth tools. Crucially, the absence of documented decision rules for exercise progression limits the transparency of the intervention. As highlighted by Hansford et al [48] and Conrado Ignacio et al [49], detailed reporting of intervention components—beyond simple dosage parameters—is essential for ensuring reproducibility and clearly defining the functional mechanisms of the prescribed therapy. Consequently, while digital platforms successfully deliver standardized content, the lack of explicit reporting on how exercises are adjusted obscures the clinical reasoning necessary to ensure safety and effectiveness throughout the rehabilitative trajectory.

Furthermore, the total lack of information regarding baseline determination and individual tailoring (0%) highlights a fundamental challenge in digital health. According to updated preparticipation health screening recommendations [50], a comprehensive evaluation of a patient’s physical status and functional capacity is the indispensable prerequisite for ensuring safety before initiating any exercise program. The inability of current apps to document how starting levels are calibrated or adjusted to individual constraints suggests a reliance on generic protocols rather than precise clinical titration. Coupled with the complete omission of adverse event reporting, these findings indicate that current apps function as digital exercise libraries rather than sophisticated therapeutic tools, falling short of the safety and individualization standards required for clinical populations.

The exercise prescription apps reviewed in this study incorporated a higher number of BCTs than reported in previous app evaluations [29,30]; the specific selection of strategies remains highly consistent with the broader mHealth literature [29,51-54]. The predominance of self-monitoring, feedback, goal setting, and action planning in our sample mirrors the most frequently identified features in recent content analyses, confirming that these specialized tools adhere to the standard design patterns observed in general physical activity interventions. However, consistent with established behavior change theory, the mere presence of multiple techniques does not ensure effective behavioral support if their selection and configuration are not aligned with the clinical requirements of long-term exercise adherence [55].

From a clinical perspective, several predefined critical BCTs for sustaining exercise behavior in unsupervised or remote settings were absent or infrequently implemented. These included graded tasks, biofeedback, prompts or cues, and self-reward—mechanisms closely associated with progression, self-regulation, and adaptive feedback over time. Their limited representation suggests that current exercise prescription apps are primarily optimized for exercise initiation and short-term compliance rather than for dynamic adjustment and sustained behavior change, which are central to clinical exercise prescription.

Conversely, several frequently implemented techniques that were not classified as critical were consistently present across all apps, including presentation of a credible source, routine review of behavioral goals, monitoring without feedback, and repeated behavioral practice. Although these techniques may have limited independent effects on long-term adherence, they may provide complementary clinical value by reinforcing professional authority, standardizing instruction, and supporting structured follow-up [56,57]. Taken together, these findings indicate that while current exercise prescription apps insufficiently implement several clinically critical BCTs, their operational strengths may still augment clinician-guided or hybrid care models, although they may not effectively function as stand-alone behavioral interventions.

In our analysis, the exercise prescription apps achieved an average MARS quality score of 3.66 (SD 0.25), consistent with prior studies by Paganini et al [58] and Simões et al [59], which reported moderate overall quality scores for physical activity apps. Paganini et al [58] found the average quality score to be 3.60, with variations across different subcategories: information averaged 3.24; engagement, 3.19; aesthetics, 3.65; and functionality, the highest at 4.35. Simões et al [59] reported a slightly higher overall MARS score of 3.88, with functionality again scoring the highest at 4.30, followed by aesthetics, information quality, and engagement.

Notably, in our study, the score for information quality was higher at 3.90, whereas the engagement score was considerably lower at 2.50. This discrepancy may reflect the more specific and task-oriented nature of our evaluated apps compared to the broader range of apps included in other studies, which might engage users differently. Despite substantial download figures, the apps we reviewed showed limited evidence of effectiveness, a finding that corroborates previous evaluations [60-62], which consistently highlighted a shortage of evidence-based content and professional guideline adherence in the mHealth landscape.

From a clinical perspective, the findings of this study indicate that currently available exercise prescription apps should not be used as stand-alone prescribing tools but rather incorporated into clinician-guided or hybrid care models. Although these apps consistently deliver structured exercise instructions and provide visibility of patient adherence, the absence of explicit progression logic, individualized adjustment, and several clinically critical BCTs limits their capacity to independently support safe and effective exercise prescription. Consequently, clinicians remain essential for determining appropriate starting levels, explaining progression criteria, and managing exercise-related risks, particularly in older adults and individuals with chronic diseases [63].

At the same time, the operational features that are consistently implemented across apps such as standardized exercise demonstrations, routine goal review, adherence tracking, and low cognitive load instruction delivery may still offer complementary clinical value. When used alongside professional judgment, these features can reduce instructional burden, support treatment consistency, and facilitate structured follow-up between clinical encounters. In this context, exercise prescription apps may function as digital extensions of clinician-led care rather than autonomous therapeutic systems.

For developers, these findings underscore the need to move beyond technically feasible features toward closer alignment with clinical exercise prescription standards. Enhancing transparency around progression rules, incorporating mechanisms for individualized adjustment, and prioritizing BCTs associated with long-term adherence may improve the clinical relevance of future applications. Most importantly, adopting a co-design approach—actively involving both clinical practitioners and end users in the development process—is essential to bridge the gap between technical implementation and real-world clinical utility [64,65].

This study has several limitations. First, search results were inherently transient and geographically specific due to localized app store algorithms and the rapidly evolving mobile market [66]. While our semantic search strategy targeted popular, clinically relevant apps rather than an exhaustive catalog, the findings represent a 2024 snapshot, and the currency of these results may diminish, as apps undergo frequent updates. Second, while the interrater reliability was high for CERT and BCT assessments, the MARS evaluation yielded relatively lower consistency, reflecting the subjective nature of aesthetic and engagement metrics; however, all discrepancies were resolved through rigorous consensus-based discussion to ensure data validity.

Third, the CERT and BCTTv1 frameworks, originally designed for research interventions, may not fully capture modern app functionalities such as automated feedback, potentially creating a structural mismatch in scoring. Fourth, we identified the presence of BCTs but did not evaluate the quality of their implementation or their actual impact on user behavior. Furthermore, the lack of longitudinal adherence and clinical outcome data limits our assessment of real-world therapeutic impact.

While our study establishes fundamental professional quality and usability benchmarks for exercise prescription apps, it serves primarily as a foundational step mapping the current landscape. Future work should build upon these findings by transitioning from quality assessment to rigorous clinical verification.

Specifically, longitudinal randomized controlled trials are needed to determine if the high-scoring, professionally vetted apps identified in this study translate to better patient adherence and health outcomes compared to traditional methods. Furthermore, considering that the effectiveness of digital exercise prescription systems may vary significantly across conditions, these investigations should target specific clinical cohorts—such as patients with cardiac conditions, neurological disorders (eg, poststroke rehabilitation), or musculoskeletal issues (eg, sarcopenia)—rather than general users.

Crucially, to match the complexity of clinical needs, efficacy trials must move beyond self-reported measures and incorporate objective physiological outcomes, such as laboratory data (eg, inflammatory markers and glycemic control) and body composition metrics assessed via bioelectrical impedance analysis. Finally, integrating wearable sensors for real-time biofeedback represents a promising frontier for ensuring that these digital prescriptions are delivered safely and effectively.

Using a clinician-centered, multidimensional evaluation framework, this study systematically assessed the clinical usability of popular, no-cost exercise prescription apps. Although these apps consistently met basic structural requirements and demonstrated strong technical functionality, they lacked essential features related to progression, individualization, intervention transparency, and sustained behavior change support. As a result, their capacity to independently deliver clinically appropriate exercise prescriptions remains limited.

Nevertheless, the standardized delivery of exercise instructions, structured translation of prescriptions into actionable tasks, and support for adherence monitoring suggest that these apps may still add value within clinician-guided or hybrid care models. Future development should prioritize a co-design approach involving both clinicians and users to bridge existing gaps. This collaborative strategy is essential to align apps with clinical standards and behavior change principles, thereby enhancing their safety, effectiveness, and clinical relevance.