Figure 1. Experimental design, univariate approach and multidimensional linked analytical workflow.

(a) Collected outcomes measures spanning information about lesion characteristics, motor, cognitive and general health domains. (b) Variables selected in an arbitrary fashion. Bar graphs reflect estimated marginal means of significant main effects and line graphs reflect significant interactions. (c) Frequency distribution and piechart of univariate p-values (d) Three experiments (Double-combo Mino, Double-combo LM11A-31 and Triple-combo) were cross-curated and merged into one master database. Outcome variables of all 202 rats are fed into a non-linear principal component analysis (NL-PCA). NL-PCA handles different analysis levels (e.g., ordinal and numeric) in the dataset by optimal-scaling transformations. The NL-PCA loading pattern shows the weight of every outcome variable on the obtained PCs. Individual subject-level PC scores are calculated by summing the optimally-transformed data variable values weighted by loadings. A linear mixed model (LMM) tested the effect of treatment group on the multidimensional outcome measure (i.e., PC score). Abbreviations: LM, LM11A-31; Mino, minocycline; PC, principal component.