Fig. 1: Cluster centers of the three-cluster sparse k-means solution described by behavioural task performance. | Schizophrenia

Fig. 1: Cluster centers of the three-cluster sparse k-means solution described by behavioural task performance.

From: Behavioural phenotypes of intrinsic motivation in schizophrenia determined by cluster analysis of objectively quantified real-world performance

Fig. 1

Cluster-wise means and bootstrapped (bias-corrected and accelerated) 95% confidence intervals are shown for the clustering input variables, derived from the Activity Preference Task (APT) and Novelty Exploration Task (NET). All variables are shown on a common scale such that higher values indicate increasing performance, with appropriate variables being reverse scaled (rev.). Based on their characteristic low NET (with medium APT) performance, low APT (with medium NET) performance, and high APT and NET performance, we respectively refer to the clusters as Low Exploration, Low Activity, and High Performance. An alternative visualization of these data for cluster-wise comparison of individual clustering input variables is provided in Supplementary Fig. S1. Task variable (sparse k-means weight): APT Active Time (0.27), duration on active engagement option; Switches (0.07), number of switches between active and passive engagement options, reversed; Active Intensity (0.25), average hand speed during periods of active engagement; Active Persistence (0.25), index of tendency to sustain continuous active engagement; NET Distance (0.09), total distance travelled; Spatial d (0.17), index of complexity of locomotion, reversed; VisObjExp Count (0.26), number of objects visually explored; VisObjExp Time (0.27), duration of visual object exploration; TacObjExp Count (0.56), number of objects physically explored; TacObjExp Time (0.55), duration of physical object exploration.

Back to article page