A low-dimensional representation of arm movements and hand grip forces in post-stroke individuals

Abstract

Characterizing post-stroke impairments in the sensorimotor control of arm and hand is essential to better understand altered mechanisms of movement generation. Herein, we used a decomposition algorithm to characterize impairments in end-effector velocity and hand grip force data collected from an instrumented functional task in 83 healthy control and 27 chronic post-stroke individuals with mild-to-moderate impairments. According to kinematic and kinetic raw data, post-stroke individuals showed reduced functional performance during all task phases. After applying the decomposition algorithm, we observed that the behavioural data from healthy controls relies on a low-dimensional representation and demonstrated that this representation is mostly preserved post-stroke. Further, it emerged that reduced functional performance post-stroke correlates to an abnormal variance distribution of the behavioural representation, except when reducing hand grip forces. This suggests that the behavioural repertoire in these post-stroke individuals is mostly preserved, thereby pointing towards therapeutic strategies that optimize movement quality and the reduction of grip forces to improve performance of daily life activities post-stroke.

Conflict of interest statement

The authors declare no competing interests.

© 2022. The Author(s).

Figures

Figure 1

Overview of the approach to capture the control of arm movements and grip forces. Kinematic and kinetic time-series were collected in a control and post-stroke population using a technology-based assessment with a goal-directed functional task, the Virtual Peg Insertion Test (representative raw data in left panel). After a pre-processing and temporal segmentation (middle panel), functional Principal Component Analysis (fPCA) was applied. fPCA allows reconstructing the time-series with a set of low dimensional basis functions, named fPCs. These fPCs can have a subject-specific shape and explain certain variance of the input signal, which was compared between a representative control and post-stroke individual, indicating a similar shape of the fPCs and a different distribution of variance explained between the subjects (right panel). The control was male and 40 years old. The post-stroke individual presented a mild upper limb sensorimotor impairment according to the Fugl-Meyer assessment for the upper extremity (score 55, age 52 years, male).

Figure 2

Movement and hand grip force coordination during the transport phase of the functional task. The preprocessed end-effector velocity (A) and grip force rate (C) signals were visualized for the paretic side of post-stroke subjects (red) and a healthy age-matched control population (gray). In addition, the shapes of the fPCs for the velocity (B) and the force rate (D) signals were visualized. The time-series of the stroke and control population were compared using statistical parametric mapping and the p- and t-values of significant periods annotated.

Figure 3

Grip force coordination during the force buildup and the force release phases of the functional task. The preprocessed end-effector grip force rate signals during buildup (A) and release (C) phases were visualized for the paretic side of post-stroke subjects (red) and a healthy age-matched control population (gray). In addition, the shapes of the fPCs for the buildup (B) and the release (D) signals were visualized. The time-series of the stroke and control population were compared using statistical parametric mapping and the P- and t-values of significant periods annotated.

Figure 4

Number of fPCs required to explain 90% of the variance in the input signal. This information was provided for each task phase and modality (end-effector velocity and grip force rate), visualized for the paretic side of post-stroke subjects (dark red), the non-paretic side of post-stroke subjects (light red), the dominant arm of the healthy age-matched control population (light gray) and the non-dominant arm of the healthy age-matched control population (dark gray). Horizontal black line indicates median, boxes represent the IQR, and the whiskers the min and max value within 1.5IQR. Horizontal solid line on top indicates results from an omnibus test and dashed lines from post-hoc tests. *Indicates P < 0.05, **P < 0.001.

Figure 5

Population-level comparison of variance explained per fPC, task phase, and modality. Horizontal solid line indicates results from an omnibus test and dashed lines from post-hoc tests. *Indicates P < 0.05, **P < 0.001.