JMIR Publications

ABSTRACT

Background:

Mobile phone addiction among children and adolescents is a growing global concern, with prevalence reaching 22%–32%. Most research has focused on eastern China or college students, and few studies have examined the heterogeneous risk profiles of children and adolescents in Yunnan Province, a multi-ethnic border region of southwestern China.

Objective:

This study aimed to characterize the prevalence of mobile phone addiction among children and adolescents in Yunnan Province, to identify distinct subgroups with different addiction profiles using k-means cluster analysis, and to describe their behavioral patterns and risk characteristics, to facilitate more targeted identification of high‑risk groups and inform future interventions and policy decisions.

Methods:

Validated questionnaires, including a sociodemographic survey and a mobile phone addiction screening scale, were used to collect data from children and adolescents in Yunnan Province selected by cluster random sampling. Non-parametric tests, chi-square tests, and multivariate binary logistic regression were used to examine associated factors. K-means cluster analysis was used to examine heterogeneity. The 16 prefectures were divided into a development set (12 prefectures) and an external validation set (4 prefectures). Within the development set, a prefecture-based cross-validation approach was adopted, with 11 prefectures as the training set and the remaining one as the validation fold in each iteration. The final clustering solution was applied to the external validation set.

Results:

A total of 44,153 children and adolescents were included. Age, sex (female vs. male), grade (Grades 5–10 vs. Grade 4), boarding status, only-child status, family size, maternal education, household income, religion, alcohol use, smoking, peer and family relationships, physical activity, and self-reported academic performance were significantly associated with mobile phone addiction. K-means clustering identified three profiles: higher-grade socially active urban students, younger day students, and ethnic minority boarding students. Cluster proportions differed by less than 1 percentage point between the development and external validation sets, and cluster-specific associations remained significant and directionally consistent in the external validation set.

Conclusions:

As shown by the K-means cluster analysis, three heterogeneous high-risk profiles of mobile phone addiction were identified and their robustness confirmed through external validation across geographically distinct prefectures, underscoring the need for subgroup differentiation for precision screening. These profiles may support efficient screening of at-risk children and adolescents and guide targeted psychological and behavioral assessment and intervention for high-risk groups.