Figure 3: Evaluation of the clusters obtained from k-means clustering. The explained variance (a) indicates how much of the variance in the data can be explained by the assignment of patients to the clusters and should therefore be maximized. The within-cluster sum of squares (WCSS, b) quantifies the deviation of the cluster members from each cluster centroid and should be minimized to obtain compact clusters with large inter-cluster distances [59]. The changes in explained variance (c) and WCSS (d) with each additional cluster for the models based on the linear slopes provide a more detailed view on the fit of each cluster number.