The global trend of population aging is increasing at an unprecedented rate. According to the latest report released by the United Nations, the global older population is expected to increase to 1.4 billion by 2030, and by 2050, the number will exceed 2.1 billion (1). Studies show that older adults are sedentary for an average of 9.4 h per day (2) and fail to meet current Physical Activity (PA) recommendations (3). The combination of prolonged sedentary behavior (Sedentary Behavior, SB) and low levels of PA results in a significant imbalance between PA and SB. This imbalance is accompanied by a significant increase in the prevalence of chronic non-communicable diseases (e.g., cardiovascular diseases, diabetes, osteoporosis, sarcopenia, and obesity) (4–11), which puts a tremendous strain on an individual’s quality of life, the burden of care on the family, and the social health-care system (12–14). Numerous research studies suggest that regular PA is the most cost-effective non-pharmacological intervention to improve health in old age, effectively reducing the risk of chronic disease, enhancing muscle function, and reducing the incidence of falls (15, 16). However, it is worrying that traditional health promotion programs (e.g., community fitness classes, paper-based promotional brochures) have significant limitations in terms of coverage, continuity, immediate feedback, and adherence management, making it difficult to provide personalized feedback (17, 18).

Currently, mobile health (mHealth) technologies, covering wearables, smartphones, tablets, mHealth apps, smartwatches, and pedometers, have gained widespread use in healthcare (19). Applying the mHealth app to exercise health management for older adults shows significant potential for enhancement: (1) provide personalized services across time and space constraints (20); (2) use sensing data (e.g., accelerometers) to enable immediate feedback and enhance self-efficacy (21); and (3) reach a wide range of people at low cost (22). However, there is a lack of systematic evaluation of the impact of stand-alone mHealth apps or apps that are part of coordinated interventions. In addition, research in this area has focused primarily on children, adolescents, and adults, with a notable lack of research in older populations. Although studies have examined the effectiveness of mHealth apps in promoting physical activity in different populations, there is significant heterogeneity in study design, intervention content, target populations, and outcome metrics, and findings vary (23–25). The study by Yerrakalva et al. (26), although it focused on older adults, was limited by the period, the limited amount of literature included, and the lack of inclusion of body composition as an outcome indicator, resulting in a lower overall quality of evidence.

In the field of research on mHealth app interventions, numerous theoretical frameworks, such as Self-Determination Theory (SDT), the Transtheoretical Model, the Health Belief Model, the Theory of Planned Behavior, and the Social Cognitive Theory (SCT), have been widely used to guide the design of relevant interventions (27). The number and type of Behavior Change Techniques (BCT) clusters are also seen as key factors influencing the effectiveness of mHealth app interventions (28). The standardized classification of BCT proposed by Migi et al. divides BCT into 16 clusters, including feedback and monitoring, rewards and threats, goals and planning, knowledge sharing, social support, and outcome comparison (29). Such categorization helps to clarify which BCT clusters are more effectively used in applications, which in turn promotes the promotion of PA and Physical Fitness (PF) (28). In summary, the effectiveness of mHealth app interventions is influenced by a variety of factors, including the type of app, intervention characteristics, theoretical paradigms, and BCT clusters. Despite the large number of reviews of mHealth app-based interventions in the existing literature, there are still significant gaps in the research.

The purpose of this study was to evaluate the effectiveness of a mHealth app-based intervention in promoting PA/moderate to vigorous physical activity (MVPA), reducing SB, and lowering body mass index (BMI) in older adults. Subgroup analyses were also conducted to explore the moderating effects of the mHealth intervention, such as the type of application, the theoretical framework, and the validity of the BCT clusters on the study’s findings. Provide targeted guidance recommendations for further advancing digital technology interventions for health promotion in older adult populations.

This study comprehensively and integrally assessed the effect of mHealth interventions on PA, MVPA, SB, and BMI through a systematic review and meta-analysis. Results showed that mHealth-based interventions effectively elevated PA: (SMD = 0.18, 95%CI: 0.01 to 0.35) and MVPA (SMD = 0.48, 95%CI: 0.20 to 0.75) in older adults, while effectively reducing SB (SMD = −0.55, 95%CI: −0.79 to −0.32). However, its improvement in BMI did not reach statistical significance (SMD = −0.27, 95%CI: −0.79 to 0.25). Subgroup analyses showed significant variability in intervention effects: commercial-based applications were more effective than research-based applications in PA promotion (MVPA: SMD = 0.70 vs. 0.30), and strategies with more than 3 BCTs showed synergistic gains in the modulation of MVPA and SB (SMD = 0.49 vs. −0.77), whereas theoretical-paradigm groups showed a specific advantage only for SB reduction (SMD = −0.77). These findings confirm the core value of digital technology in empowering the health of the older adult, but also reveal the complexity and boundary conditions of the intervention effects.

In terms of facilitation mechanisms for PA and MVPA, this study found that commercial-based applications were significantly more effective than research-based applications. This phenomenon may stem from the fact that commercial products focus more on user experience design, e.g., enhancing motivation to participate through gamification elements (e.g., virtual medals, step leaderboards), whereas research applications tend to focus on functional completeness at the expense of interactive experience (23, 50). The results challenge the conventional wisdom that evidence-based design is necessarily superior to commercial products and suggest the critical role of practical design in the older population (23, 51). Of course, this could be attributed to the fact that mHealth, which is research-based, is still in the experimental stage, and its personalized design may not yet be fully developed (52).

The moderating effects of theoretical paradigms showed specificity (53). Interventions constructed based on SDT or SCT were particularly effective in reducing SB, possibly because theoretical frameworks are more adept at explaining intrinsically motivated-driven behavioral changes (e.g., autonomous choice of activity break rhythms) (54); However, their promotion of PA/MVPA was not consistent, suggesting that high-intensity activity may require more direct external incentives (e.g., stage-based rewards) (55). This finding provides an important rationale for the theoretical fit of future intervention designs.

Notably, the number of BCTs integrated significantly affected intervention efficacy: interventions that included more than three BCTs (e.g., goal-setting, real-time feedback, social support) showed a significant advantage in improving MVPA and reducing SB, confirming the “synergistic efficacy of BCTs” hypothesis proposed by Michie et al. (29). It has been shown that clusters of multiple BCTs can build self-regulating neural circuits and that goal setting can activate the dorsolateral prefrontal cortex (DLPFC) to enhance executive function and prospective memory (56, 57); Real-time feedback can enhance the computation of reward prediction errors and enhance the perception of reward for effort through basal nucleus dopaminergic projections (58); Social support, in turn, stimulates empathic motivation via the mirror neuron system (59). When these techniques cover the whole cycle of “goal formation-action monitoring-feedback regulation-strategy updating,” they can enhance the self-efficacy of older adults (56). Conversely, fragmented BCTs (≤3 BCTs) may result in a nonsignificant MVPA effect size due to failure to establish neurofunctional coupling (56).

It is thought-provoking that despite significant improvements in PA and SB, BMI did not show a synchronized decrease. This phenomenon may be attributed to the insufficient number of long-term intervention cycles included in the current study, resulting in a slow rate of BMI reduction. Improvements in PA can be achieved in a relatively short period, whereas reductions in BMI may take a longer time to show changes (60). Secondly, the increase in MVPA may not yet have reached the intensity threshold that has a significant impact on body fat metabolism. Therefore, future research should focus on two key areas: first, extending the intervention cycle to verify long-term effects (given that most existing research cycles are shorter than 12 weeks); second, adjusting changes in intervention intensity, especially mobile health (mHealth) interventions aimed at muscle strengthening (the proportion of such current studies is less than 10%) (61).

In this study, a large amount of credible and relevant literature was collected by searching Embase, PubMed, Web of Science, and Cochrane databases. The PRISMA guidelines were strictly followed during the study, and a two-person independent screening process was used to minimize selection bias. The current study is the first systematic review and meta-analysis to comprehensively include RCTs to assess the effects of mHealth interventions on PA, MVPA, SB, and BMI in older adults. In subgroup analyses, this study is the first to explore the moderators of mHealth intervention effects, including the type of application, theoretical paradigm, and BCT clusters on various indicators in older adults. Therefore, this study is highly robust and innovative.

Inevitably, this study still has certain limitations. First, since some primary studies did not specify the exact names of the intervention apps, subgroup analyses could not be conducted within individual app types. Second, although the PRISMA guidelines were strictly followed, the limited number of primary studies means that certain subgroup-analysis results may still be heterogeneous. Finally, because the outcome measures used in the primary studies were not reported in identical units, these discrepancies could bias the findings and potentially limit the effect sizes.

This study revealed that the mHealth intervention had a significant effect in elevating PA and MVPA levels in older adults and was effective in reducing SB, but did not reach statistical significance in terms of BMI improvement. Through subgroup analyses, this study further found that commercial applications showed greater advantages in promoting PA/MVPA; Integration of more than three BCTs can synergistically contribute to increased MVPA levels and reduced SB. Based on these findings, it is recommended to prioritize the use of established commercial applications that incorporate multiple BCTs in health coaching practices. Future studies should incorporate trials with a longer period. Meanwhile, it is recommended that “intervention tools” and “measurement tools” be separated, and validated independent devices be used to evaluate the results so as to reduce bias.