As smartphones become increasingly prevalent in adolescents' learning and social lives, the question of whether problematic smartphone use (PSU) itself leads to measurable psychological or behavioral dysfunction remains inconclusive. A multi-model cluster analysis was conducted based on cross-sectional data from 5th-9th graders (N=16,052; Mage=13.24, SD=1.14) in a district of Beijing. The analysis used four PSU dimensions (life interference, virtual life orientation, withdrawal and tolerance) and six psychological indicators (subjective well-being, resilience, prosocial behavior, depression, anxiety and impulsivity). After identifying distinct subgroups, SHAP attribution analysis was applied to 24 variables encompassing academic performance, individual psychology, smartphone use, and contexts relating to family and school, to identify key differentiating features. Cluster analysis based on multidimensional indicators categorized adolescents into the following five groups: Well Functioning-High PSU, Stable-Moderate, Flourishing-Low PSU, Distressed-Low PSU, and Severely Distressed-High PSU. Feature attribution analyses, with the Well Functioning-High PSU group as the reference category, revealed the features that most clearly distinguished the Severely Distressed-High PSU group in terms of psychological functioning (emotion regulation, school belonging, study interest, and growth initiative). Compared to the Flourishing-Low PSU group, key differentiating features were observed in behavioral/time management variables (study habits and weekend screen time). These findings suggest that dysfunction is more closely linked to self-regulation and social connectedness, while PSU levels are more strongly correlated with habitual behavioral and rhythmic patterns. This provides empirical evidence to support the precise identification of individuals and the development of targeted interventions.