Fig. 2: Cluster analysis result for one bipolar patient.
From: Smartphone-based markers of social connectivity in schizophrenia and bipolar disorder
A The hierarchical and partitional clustering analysis agree with the choice of three clusters. Loneliness clusters separately, and social network degree clusters with self-report of social activity. Social network reciprocity measures clusters with call and text duration. B When considering mobile communication data involving only “important contacts” the hierarchical and partitional clustering analysis agrees in the choice of 2 clusters. Loneliness clusters with mobile communication features, separately from self-report of social activity.
