This study was a scoping review that adhered to the PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews) guidelines [46]. The completed PRISMA-ScR checklist is provided in Multimedia Appendix 1. An a priori protocol was developed and registered on PROSPERO (CRD42024530851).

This review was initially registered with PROSPERO as a systematic review. Upon further consideration, we adopted a scoping review approach, as our aim was to explore and synthesize existing evidence on the use of digital technologies in AIS exercise rehabilitation and to identify key features and research gaps. A scoping review is particularly suitable for emerging topics and provides a broad conceptual overview of existing evidence [47]. Accordingly, we adapted the search strategy, data extraction, and analysis methods to align with the objectives of this scoping review. The inclusion and exclusion criteria for this scoping review are provided in Textbox 1.

Searches were conducted from the inception of the databases up to April 4, 2024, across the following five electronic databases: PubMed (National Library of Medicine), Web of Science (Clarivate Analytics), Embase (Elsevier), Cochrane Library (Wiley), and CINAHL (EBSCO), without any restrictions. The electronic search strategy (Multimedia Appendix 2) was developed and executed under the guidance of an experienced librarian. The search terms (Table 1) covered three main topics and their subtopics: AIS, exercise therapy, and digital intervention. Additional sources were identified through manual searches, including reviewing the reference lists of studies included in this scoping review to minimize the chances of missing relevant studies.

All electronic search records were stored in EndNote X9 (Clarivate) and deduplicated. Two authors (YL and FC) independently screened an initial set of 30 studies and compared their results to ensure a consistent understanding of the screening criteria. Subsequently, they independently screened the remaining records based on titles, abstracts, and full texts. Any discrepancies were resolved through discussion with a third author (ZR). Reasons for full-text exclusions were documented. Additionally, we manually screened the reference sections of eligible full-text sources to identify further relevant studies.

Data charting was collaboratively completed by two authors (YL and FC) using Excel (Microsoft Corp) and was pilot-tested and calibrated within the team. The data extraction table was developed based on the TIDieR (Template for Intervention Description and Replication) checklist [48]. Additionally, we referred to the data charting tool developed by Gooch et al [16], which extends the TIDieR checklist by incorporating elements specific to digital interventions and behavior change methods, aligning with this review. Data coding was independently performed by two authors (YL and FC), with discrepancies resolved through discussion. If consensus could not be reached, a third author (ZR) acted as an adjudicator. Quantitative data extracted from the included studies were coded into predefined categories, and qualitative author statements were also coded. The classification of digital technologies was initially based on prior reviews [16,49] and was expanded as new types of technologies were identified during data extraction. Conceptually related technologies were further grouped into broader thematic clusters. For example, “videoconferencing,” “instructional videos,” and “mobile applications” were categorized under the thematic cluster “Digital Content and Platforms.” Other charted variables (eg, intervention location and duration) were also synthesized into thematic categories by analyzing associations among the initial codes and summarizing them based on the most frequently reported features.

For extracting behavior change strategies in digital interventions, we first coded each study using the BCTTv1 and recorded the frequency of BCT use [35]. Next, based on expert consensus from prior studies linking BCTs to TDF domains in health interventions, and team consensus, the BCTs were mapped onto the TDF mechanisms of action [41,50,51]. Finally, the TDF domains were linked to the COM-B components. The Behavior Change Wheel is a guide for designing interventions, providing explicit connections between TDF domains and COM-B components [45].

Data items related to digital technologies and behavior change strategies are summarized in Table 2. The complete data extraction table is provided in Multimedia Appendix 3 [12,17,24,25,52-68]. In studies where digital technologies were integrated into broader interventions, it was sometimes difficult to isolate the effects of individual components. Consequently, behavior change strategies were coded for the entire intervention.

The quality of the included studies was assessed using the Mixed Methods Appraisal Tool (MMAT) [69,70]. The MMAT is a validated tool that enables the evaluation of the methodological quality of five study categories: qualitative research, randomized controlled trials (RCTs), nonrandomized studies, quantitative descriptive studies, and mixed methods studies. In addition to two general screening questions applicable to all study types, each study design includes five specific questions relevant to its methodology, each scored as “yes,” “no,” or “cannot tell.”

Two authors (YL and FC) independently assessed all studies. Discrepancies were resolved through discussion, and a third author (ZR) was consulted when consensus could not be reached. In line with the aim of this scoping review, no studies were excluded based on the quality assessment results [71].

Quantitative data were synthesized in Excel using descriptive statistics (eg, frequencies) and presented in tables and figures. Qualitative data on user experiences with digital interventions, including direct user quotes and authors’ interpretations, were synthesized narratively into common themes. Two authors (YL and FC) independently extracted the qualitative data and conducted open coding. Initial codes were discussed and grouped into broader categories, eventually resulting in four main themes. These themes were refined through team discussions to ensure reliability and consensus. Meanwhile, a summary and descriptive analysis were conducted on the general characteristics of included studies, digital technology characteristics, user experiences, and behavior change strategies. To further analyze the use of BCTs across studies and compare them with other reviews, BCTs were classified into 16 BCT clusters for analysis [16,72].

A total of 3267 studies were identified, including 2754 from databases and 513 from other sources. After removing duplicates and screening titles, abstracts, and full texts, 21 (0.64%) studies were included in the review (Figure 1).

This scoping review summarizes the current literature on digital exercise interventions for patients with AIS, identifying the BCTs used and their potential mechanisms of action. A total of 21 studies were included, with most (15/21, 71%) published in the last 4 years, and nearly half (9/21, 43%) being feasibility or pilot studies. These findings suggest that this is a rapidly growing field with significant potential for future research.

The included studies varied in the exercise methods. While PSSE is widely recognized for scoliosis treatment, only half of the studies used PSSE methods, with the rest favoring GTE. This diversity may result from the current lack of high-quality evidence supporting the effectiveness of either PSSE or GTE for AIS, or demonstrating the superiority of one over the other [9,73]. Additionally, no clear relationship between exercise methods and digital technologies was found, as the choice of digital tools did not seem influenced by the type of exercises. Future research should explore the effectiveness of combining specific exercise methods with digital technologies and clarify their supportive role.

Digital technologies play several roles in AIS exercise rehabilitation, each offering distinct advantages and limitations. First, digital content and platforms have transformed traditional methods into telerehabilitation through internet communication, extending the reach of face-to-face rehabilitation. The rise of telerehabilitation, likely accelerated by the COVID-19 pandemic, offers low cost and broad accessibility, particularly benefiting patients in remote areas [49]. These technologies enable physiotherapists to supervise and guide home exercises either synchronously (eg, via videoconferencing) or asynchronously (eg, via instructional videos), thereby improving the feasibility of home rehabilitation [67]. Telerehabilitation has been shown to be a safe and effective complement to AIS treatment [67]. However, it also presents challenges, such as the absence of physical contact between physiotherapists and patients and reliance on stable internet connections [67,74]. Compared to fully remote programs, several studies suggest that a hybrid model combining web-based and offline approaches may better meet patient needs. For example, Dursun et al [68] found that continuing home rehabilitation after a clinical program enhances both its effectiveness and sustainability. Second, VR and AR have been applied as auxiliary tools in conventional rehabilitation but remain underexplored in exercise-based rehabilitation [75,76]. VR creates immersive simulated environments, while AR overlays virtual elements onto real-world scenes [77]. In the included studies, VR and AR were used alongside other devices, such as a handheld device [64], a force platform [62], or a mechatronic horse-riding simulator [59]. These technologies enhance immersion, maintain attention, and boost engagement [49,75]. For instance, Rösner et al [59] found that most participants preferred immersive VR environments, suggesting that VR systems could improve user experience. However, VR and AR systems tend to be expensive, limiting their use outside of research settings, and may cause side effects such as dizziness or visual fatigue [49,77]. Future studies should focus on developing cost-effective solutions and evaluating long-term effectiveness in home rehabilitation [76].

Third, exergames have shown benefits in exercise rehabilitation for conditions such as neurological disorders and childhood obesity [55,78]. Unlike commercial games, the exergames in the included studies were custom-developed and incorporated inertial measurement unit sensors [61] or game controllers [53,55] to capture real-time postural movements. Exergames complement traditional exercise rehabilitation by providing enjoyable, gamified experiences [78,79]. Their customized design also enables researchers to tailor content to individual needs [55,79]. Nonetheless, some studies reported challenges such as declining motivation, poor adherence, and an increased risk of adverse events [55,80]. For example, Wibmer et al [55] found that patients with AIS reduced their exercise frequency as interest in exergames waned, a finding consistent with other studies [81,82]. Finally, wearable sensors monitor posture to support independent exercise, partially replacing physiotherapists and enabling more complex interventions. Current studies focus on the technical aspects of posture monitoring, consistent with previous findings that physiotherapists consider body position tracking via digital tools a key priority for home rehabilitation [26]. Wearable sensors are well-suited for home use, offering real-time feedback and allowing physiotherapists to adjust programs remotely based on collected data, thereby facilitating personalized rehabilitation [83,84]. However, limitations include concerns about data privacy and limited adaptability in home environments [85]. Although wearable sensors have been widely studied in general telerehabilitation, this review highlights their underutilization in AIS rehabilitation, suggesting a promising direction for future research. Future studies should explore the integration of multiple digital technologies to develop more effective, comprehensive rehabilitation programs.

Current digital interventions lack sufficient focus on feedback design, despite the crucial role of multimodal feedback in helping patients accurately understand and perform movements [86]. Visual and haptic feedback, in particular, warrant further research. VR and AR technologies provide immersive visual feedback by overlaying virtual images onto real environments or creating fully virtual settings, which may reduce patients’ perception of physical pain during corrective exercises [86,87]. Haptic feedback offers direct and precise motion guidance, which is especially valuable in home-based rehabilitation without physiotherapist involvement. Andrade et al [67] noted that telerehabilitation lacks the direct physical contact of traditional therapy, limiting hands-on treatment by physiotherapists. Haptic feedback could address this limitation by guiding patients to control limb positions accurately, compensating for the shortcomings of telerehabilitation. Future research should delve deeper into feedback modality design in digital interventions to improve rehabilitation outcomes and patient experiences.

Our findings show that “Shaping Knowledge,” “Goals and Planning,” “Feedback and Monitoring,” and “Repetition and Substitution” are common BCT clusters, consistent with broader reviews [16,88]. Additionally, “Social Support” and “Antecedents” are frequently used, likely due to the need for auxiliary equipment (eg, yoga mats and resistance bands) and digital devices in scoliosis therapy, as well as the social support required to help patients cope with challenges [26]. Although “Reward and Threat” and “Natural Consequences” were rarely used, they have shown potential in other studies, such as interventions for musculoskeletal and pediatric patients [88,89]. A review of digital interventions for children and adolescents also highlights the role of “Reward Systems” and “Guidance” in enhancing cognition and motivating behavior change [90]. Despite identifying common BCTs, there is still insufficient evidence on the most effective ones and how to integrate them optimally, which requires further research [89].

This review identifies “Knowledge,” “Skills,” “Environmental Context and Resources,” “Social Influences,” and “Goals” as the primary TDF domains that may support exercise rehabilitation and influence patient behavior, consistent with previous studies highlighting their positive impact [91,92]. Mapping to the COM-B model reveals that most studies focus on “Capability” and “Opportunity,” with “Psychological Capability” and “Physical Capability” being the most frequently mentioned subcomponents. This is expected since exercise rehabilitation aims to improve health through physical activity and exercise, and “Capability” is essential for behavior to occur [93]. Successful exercise rehabilitation for patients with AIS relies on their understanding of exercises, ability to use digital tools, and physical skills. Additionally, “Physical Opportunity” and “Social Opportunity,” as external drivers of behavior, are widely mentioned. Digital tools create opportunities for exercise and social interaction by optimizing environments and facilitating remote interactions between physiotherapists, patients, and parents throughout the rehabilitation cycle [94,95].

Although “Motivation” is recognized as a key factor in rehabilitation, current research gives it insufficient attention, particularly regarding “Automatic Motivation,” which is rarely discussed [96]. Most studies focus on “Reflective Motivation” as an external driver, primarily using the BCT “Action Planning” to raise patients’ awareness of rehabilitation plans without exploring additional strategies. “Automatic Motivation,” an unconscious, intrinsic driving force, is crucial for long-term adherence, as it fosters autonomous patient engagement [93,97,98]. Although some studies have used digital technologies to increase engagement and stimulate enthusiasm, this often leads to only temporary motivation, as the novelty of the technology tends to diminish over time [55,81,82]. Previous research has begun to explore strategies, such as BCTs, to enhance automatic motivation. A qualitative study on survivors of stroke found that BCTs related to reward or incentive, such as material and social rewards, can encourage sustained physical activity by activating automatic motivational processes, which align with the TDF domain of Reinforcement [99]. Additionally, studies based on self-determination theory suggest that certain self-regulatory BCTs, such as self-monitoring and reviewing behavioral goals, may enhance intrinsic motivation for physical activity; however, their specific connection to automatic motivation remains unclear [100,101]. Therefore, future studies should further investigate how digital interventions can effectively strengthen automatic motivation through these strategies to support lasting behavioral change in patients with AIS.

Based on these findings, we recommend the following key features for developing digital exercise interventions to empower patients with AIS in rehabilitation.

To our knowledge, this scoping review is the first to summarize the use of digital technologies in AIS exercise rehabilitation, highlighting a promising area for further research. Additionally, it is the first to identify and map BCTs in digital interventions to TDF domains and COM-B model components, providing insights into the theoretical foundations of these interventions. This approach bridges the gap between theory and practice, offering valuable guidance for health care professionals and researchers in developing effective digital interventions.

However, this review has several limitations. First, none of the studies explicitly reported the BCTs used, so BCT identification relied on intervention descriptions using the BCTTv1 and expert consensus [35,41,50,51]. This process may introduce subjectivity, potentially affecting the results. To mitigate this, two researchers independently coded the data and held regular discussions. Second, the broad inclusion criteria encompassed studies involving other types of patients with scoliosis, patients with AIS complications, or nontarget populations. While this approach increased the number and diversity of studies analyzed, it also introduced heterogeneity in the study populations, which may limit the generalizability and clinical applicability of the findings to typical adolescent patients with AIS. Third, the varied roles of digital technologies added coding complexity, making it difficult to isolate their independent effects from other intervention components. Fourth, more than half of the included studies were feasibility or pilot studies, study protocols, or case studies, representing research types characterized by limited methodological rigor, incomplete intervention outcomes, or low generalizability. This prevalence inevitably reduces the overall robustness and applicability of the synthesized findings. Additionally, restricting the review to English-language studies may have affected its comprehensiveness. Finally, we were unable to retrieve the full text of 15 potentially eligible studies, as they were not available in databases or web-based sources. The exclusion of these studies may have introduced selection bias.

Future research should follow standardized reporting guidelines, such as CONSORT (Consolidated Standards of Reporting Trials) and TIDieR, to ensure detailed descriptions of digital interventions, including the content and implementation procedures. Additionally, there is a need to establish consistent definitions and classification frameworks for digital technologies to standardize the reporting of digital components across studies. This will enhance transparency, reproducibility, and comparability between interventions [48,102]. Furthermore, an initial theoretical understanding is essential for the design and evaluation of complex interventions, particularly those targeting behavior change [23,103]. However, none of the included studies reported a theoretical basis. Future research should explicitly report behavior change strategies, including the underlying theories and common taxonomies (eg, BCTTv1), to clarify intervention mechanisms and support the development of effective digital interventions [16,104]. Finally, a more comprehensive evaluation of digital interventions is needed. Future studies should incorporate long-term follow-up and adopt standardized outcome measures (eg, Cobb angle) to assess effectiveness. In addition, more in-depth qualitative research is required to explore user experiences, assess usability and home applicability, and inform future intervention design.

Additionally, due to limited evidence regarding how digital tools are practically used in real-world settings, designing and implementing digital interventions that effectively maintain adherence remains challenging. This review offers actionable implications for clinical practice. Physiotherapists and clinicians may consider using digital technologies as a complement to traditional interventions. When selecting appropriate technologies, it is important to account for their advantages and limitations, rehabilitation goals, and patient characteristics. For home-based rehabilitation, integrating digital interventions with regular in-person follow-ups is recommended. Moreover, incorporating behavior change elements into the intervention can help achieve desired target behaviors, such as enhancing engagement through goal setting. Finally, this review highlights key factors for empowering patients in digital interventions. These factors are linked to enhancing motivation for exercise adherence and can thus inform the design of conventional interventions as well.

Digital technologies hold significant potential in AIS exercise rehabilitation, transforming traditional approaches by enhancing accessibility, patient engagement, and personalization. However, current interventions lack a strong theoretical foundation and explicit behavior change strategies, particularly in sustaining long-term patient motivation. To fully realize the benefits of digital interventions, future research should integrate behavior change theories and techniques to develop evidence-based, theory-driven digital solutions. These interventions should provide effective and sustainable rehabilitation options, improve long-term outcomes, and enhance the user experience.