Rapid digitalization and population aging have made health management for urban sub-healthy older adults a critical priority (1, 2). As per the 56th Statistical Report on Internet Development of China issued in July 2025, China has now conquered 161 million internet users of 60 years and above. This represents a penetration rate of 52.04%. Mobile Phone Dependence (MPD) is becoming pervasive within this demographic (3). MPD entails a loss of behavioral control resulting from prolonged, excessive usage. It triggers intense cravings and subsequently impairs social function. Sub-healthy older adults already face physiological and psychological decline. Therefore, they are particularly vulnerable to this negative cycle of MPD (4).

Sub-health in older adults is a transitional “third state” between health and disease (5). It is characterized by a significant decline in physiological function, mental health, and social adaptation, yet it does not meet the clinical diagnostic criteria for a specific illness (6). Epidemiological data shows that sub-health is highly prevalent among China’s urban older adults, with approximately 55–70% of people aged 60 and older experiencing various sub-health symptoms (7). Regarding diagnostic standards, the Sub-Health Measurement Scale (SHMS V1.0) is a widely used tool for standardized assessment (8). This scale effectively identifies issues such as physical fatigue, sleep disturbances, and cognitive decline (9). For the urban sub-healthy older adults, the combination of sedentary lifestyles and the complexities of digital mobile phone use has created an urgent need for health management (10). Targeted interventions are necessary to prevent the deterioration of physical, mental, and social functions into diagnosable clinical diseases.

Self-control (SC) is the fundamental ability to regulate individual behavior with societal expectations. Generally, SC exhibits a negative correlation with MPD at the trait level (11). However, this relationship is not merely additive. It creates a vicious cycle of “resource depletion, diminished control, intensified dependence.” Individuals expend self-control resources to resist digital temptation. Once these resources are exhausted (12). Existing research predominantly focuses on linear correlations. This approach ignores intra-population heterogeneity. Significant disparities likely exist in how MPD and SC combine within different older cohorts. Identifying these variations is essential for understanding their Tai Chi exercise and health behaviors.

Tai Chi serves as a traditional Chinese health-preserving exercise with distinct advantages for sub-healthy older adults (13, 14). Physiologically, its gentle movements enhance muscle strength, flexibility and balance. It also effectively relieves physical fatigue (15, 16). Psychologically, the discipline focusing on breath and movement helps clear distracting thoughts. This fosters emotional stability and psychological resilience (17, 18). Socially, Tai Chi communities provide vital support networks (19, 20). Previous studies suggest links between physical activity, SC, and MPD. Generally, high SC motivates activity, while high MPD reduces it. However, these variable-centered studies fail to capture subtype differences within the “MPD-SC” dimension (21).

LPA serves as an individual-centered analytical method to address this gap. It identifies latent subgroups with similar patterns, overcoming the limitations of variable-centered approaches (22). To our knowledge, no previous study has applied LPA to categorize MPD and SC among sub-healthy urban older adults. Nor have studies examined the specific association with Tai Chi. The study offers three key innovations. First, it uses LPA to uncover heterogeneous subtypes. This addresses the previous neglect of intra-population variation. Second, it quantifies differences in Tai Chi participation across these distinct subtypes. This supports the development of precision intervention strategies. Third, it focuses on the unique interactions between MPD, SC, and health behaviors in this vulnerable population. Therefore, the findings hold significant practical value.

In summary, this study utilizes LPA to dissect latent categories of MPD and SC among sub-healthy urban older adults. It further explores the relationship between these categories and Tai Chi exercise. This not only enriches theoretical research on MPD, SC, and health behaviors. It also provides actionable evidence for intervention programs. Key goals include preventing MPD, enhancing SC, and promoting Tai Chi. Ultimately, this research aids in improving the quality of life for this population and facilitating healthy ageing.

The findings reveal that mobile phone dependency and self-control among sub-healthy older adults comprise four distinct types: Low dependency-Medium control, High dependency-No control, No dependency-High control, and Moderate dependency-Low control. Notably, High dependency-No control category accounted for 37.0% of all participants. The No dependency-High control category comprised 34.1%. This proportion is significantly higher than the other two categories. They also surpass findings from previous studies (36). Classification outcomes differ from prior research as well (37). This discrepancy likely stems from the simultaneous consideration of two variables in this study. This method produces a more granular segmentation of the latent profiles. The high prevalence of dependency indicates that the sub-healthy urban older adults cohort generally struggles with low self-control. They also exhibit poor health habit formation. This is closely associated with physiological and psychological changes characteristic of sub-health in old age (38). From a neurobiological perspective, the sensitivity of the reward system in the brains of urban sub-healthy older adults is often diminished. Consequently, they require prolonged mobile phone usage to achieve the same level of pleasure. This leads to a loss of control (39). The distribution of these profiles reveals group-specific demographic patterns. These insights are crucial for understanding intrinsic heterogeneity within this population. They also aid in developing precise intervention strategies.

On the gender dimension, older women have much higher proportions of “Moderate dependency-Low control” groups than men. This discrepancy may be related to women’s more nuanced and intense intrinsic needs for emotional expression and social connection in later life (40). Research shows that women are more likely to experience social circle contraction post retirement, leading them to rely on mobile phones to keep social interaction and emotional support. If self-regulation abilities do not adapt synchronously, this leads to heightened susceptibility to acquiring mobile phone dependency (41). This also suggests that when promoting group-based mind–body exercises such as Tai Chi, emphasis should be placed on their role in fulfilling and guiding the socio-emotional needs of older women.

Regarding age, individuals aged 70–79 comprise the largest proportion of the “High dependency-No control” group. This likely reflects a transitional phase in physical and mental function (42). They still have a certain willingness to explore new technologies like smartphones. However, partial physical decline increases their reliance on devices for accessing daily services. Additionally, the natural deterioration of the prefrontal cortex-linked to self-control-often occurs during this stage. Collectively, these factors make this subgroup the most vulnerable to dependency (43). Conversely, those aged 80 and above predominantly fall into the “non-dependent-high control” or “moderately dependent-low control” categories. This is attributed to fixed lifestyles and lower mobile phone acceptance. They also retain established behavioral self-regulation mechanisms (44). However, some older adults may develop dependency in specific scenarios due to health management needs, but without having systematic control (45).

On educational attainment, those with master’s degrees or higher present a significantly higher proportion in the “Moderate dependency-Low control” group. This reveals a contradictory pattern of “high cognition, low execution.” Advanced digital literacy facilitates extensive mobile utilization for academic exchange and smart health management. However, specific knowledge structures and social attributes may cause a cognitive-behavioral disconnect (46). Consequently, interventions for highly educated groups must extend beyond knowledge dissemination. The focus should shift to cultivating sound behavioral habits.

Regarding income, high-earners primarily populate the “High dependency-No control” and “No dependency-High control” groups. This suggests greater access to alternative recreational resources and premium wellness facilities. Objectively, this provides better conditions for self-regulation (47). However, professional or social continuity requirements may induce specific dependencies in some high earners (48, 49).

In summary, significant intra-group heterogeneity exists regarding mobile phone dependency and self-control among urban sub-healthy older adults. This diversity correlates strongly with gender, developmental stages, education, and socioeconomic resources. Future promotion of traditional exercises like Tai Chi must account for these variations. Interventions should respect the specific psychological traits and motivations of different subgroups. Implementing targeted facilitation strategies could improve physical and mental health while bolstering smartphone usage regulation.

The findings indicate a strong association between Tai Chi practice and the latent profiles of MPD and SC. Specifically, individuals with lower Tai Chi exercise frequency and shorter durations were more likely to be categorized into the “Moderate dependency-Low control” and “High dependency-No control” groups. Conversely, higher intensity and regularity of Tai Chi practice were significantly correlated with the “No dependency-High control” category. This suggests that Tai Chi activity serves as a significant correlate of these psychological characteristics. Older adults with less physical activity exhibit higher dependency and reduced self-regulation, which may be related to weaker psychological adaptation and limited self-regulatory resources (50). In contrast, regular practitioners demonstrate higher self-control and lower dependency tendencies, supporting the link between Tai Chi and enhanced self-regulatory capacity. Consistent with existing research, these findings suggest Tai Chi’s feasibility as a mind–body intervention (51). It is associated with mitigated excessive mobile use through the potential bolstering of self-control capabilities. Furthermore, long-term practice is linked to improved physical health, self-efficacy, and life satisfaction (52). Therefore, developing differentiated promotion strategies is essential. Tailoring interventions to specific MPD-SC profiles may help enhance self-regulation, reduce dependency risks, and facilitate a transition toward healthier states.

While the odds ratio for Tai Chi practice in certain categories remains high (OR = 111.599), this primarily reflects the exceptionally strong protective association between regular exercise and the “no dependency-high control” profile, rather than statistical artifacts. However, due to the relatively small sample size in specific subgroups, these magnitude estimates should be interpreted as indicative of a strong directional effect rather than a precise point estimate.

In addition, the current study qualitatively explored the potential moderating effects of gender and age on the relationship between Tai Chi and the identified latent profiles. The preliminary findings suggest that Tai Chi exercise may exert a universally positive impact regardless of gender or age within the sub-healthy older adults. However, due to the limited sample size in certain specific subgroups (e.g., C4), this study was not powered to detect complex interaction effects. Future research with larger, more diverse cohorts is warranted to conduct robust subgroup analyses and interaction testing.

This study contains several limitations. Firstly, the cross-sectional design precludes causal inferences regarding MPD, self-control, and Tai Chi practice. Future longitudinal studies are necessary to address this gap. Second, the sample was restricted to sub-healthy residents of Xuzhou City, Jiangsu Province. This may introduce regional bias. Future research requires larger sample sizes and broader geographical coverage. Third, the study did not examine disease heterogeneity within the sub-healthy population. Future interventions should account for variables such as disease type, severity, medical expenses, and recurrence rates. Finally, the survey was limited to individuals aged 60 and above. Future stratified research in various age groups is likely to determine the turning points of dependency and self-control. This would facilitate targeted preventive measures at various life stages to optimize health status.

Moreover, One significant limitation is the potential selection bias, as all participants were recruited from established Tai Chi practice sites. This active population may already possess higher self-control or lower levels of mobile phone dependence than sedentary older adults. Consequently, the observed associations may be overestimated. Future studies should include a sedentary control group to further validate these findings.

Latent profile analysis (LPA) was employed in this study to determine four different subgroups among urban sub-healthy older adults. It revealed significant disparities in MPD and self-control levels within these groups. Demographic factors such as gender, age, ethnicity, marital status, educational attainment, and monthly income are key predictors of category membership. The findings confirm that Tai Chi, as a traditional mind–body exercise, fosters physical and mental well-being in this population. However, its efficacy varies across different subgroups. This discovery contributes to population-specific health behavior theory. It also provides empirical evidence for designing multi-tiered, individualized health promotion programs. Accurately identifying group characteristics allows for the precise utilization of Tai Chi. Ultimately, this approach improves sub-health status and strengthens behavioral regulation capabilities.