Figure 1.

A schematic illustration of the tensor data analysis pipeline implemented in this study. (A) The 5-way tensor data $\underset{\_}{\mathbf{X}}$ (Spatial$\times$Temporal$\times$Object$\times$Participants$\times$Trials); (B) The NCP decomposition $\underset{\_}{\mathbf{X}}\sim \mathbf{A}\circ \mathbf{B}\circ \mathbf{C}\circ \mathbf{D}\circ \mathbf{E}$; (C) The first mode of the estimated factor matrices including a spatial (muscle and force) component, a temporal component, an object (heavy or light) component, a participant component and a trial component; (D) Using the participant component to predict clinical test values (e.g., manual dexterity) or participant characteristics (e.g., age) using regression and classification analyses.