Fig. 3From: Geometric characterisation of disease modulesDistance-based clustering of disease modules. ncMCE, a nonlinear dimensionality reduction algorithm, was applied to the matrix of pairwise hyperbolic distance-based DM separations. The resulting two-dimensional projection of the data separates DMs according to the disease type reported in the ICD-10. Some interesting DMs that were not correctly clustered are highlighted. The concordance score (C-score) over Dim 2 for this projection is 0.758. The points corresponding to endometriosis and down syndrome were slightly moved to make them visibleBack to article page