In-depth quantification of bimanual coordination using the Kinarm exoskeleton robot in children with unilateral cerebral palsy

Abstract Background Robots have been proposed as tools to measure bimanual coordination in children with unilateral cerebral palsy (uCP). However, previous research only examined one task and clinical interpretation remains challenging due to the large amount of generated data. This cross-sectional study aims to examine bimanual coordination by using multiple bimanual robotics tasks in children with uCP, and their relation to task execution and unimanual performance. Methods The Kinarm exoskeleton robot was used in 50 children with uCP (mean age: 11 years 11 months ± 2 years 10 months, Manual Ability Classification system (MACS-levels: l = 27, ll = 16, lll = 7)) and 50 individually matched typically developing children (TDC). All participants performed three tasks: object-hit (hit falling balls), ball-on-bar (balance a ball on a bar while moving to a target) and circuit task (move a cursor along a circuit by making horizontal and vertical motions with their right and left hand, respectively). Bimanual parameters provided information about bimanual coupling and interlimb differences. Differences between groups and MACS-levels were investigated using ANCOVA with age as covariate (α < 0.05, $${\upeta }_{p}^{2}$$ η p 2 ). Correlation analysis (r) linked bimanual coordination to task execution and unimanual parameters. Results Children with uCP exhibited worse bimanual coordination compared to TDC in all tasks (p ≤ 0.05, $${\eta }_{p}^{2}$$ η p 2 = 0.05–0.34). The ball-on-bar task displayed high effect size differences between groups in both bimanual coupling and interlimb differences (p < 0.001, $${\eta }_{p}^{2}$$ η p 2 = 0.18–0.36), while the object-hit task exhibited variations in interlimb differences (p < 0.001, $${\eta }_{p}^{2}$$ η p 2 = 0.22–0.34) and the circuit task in bimanual coupling (p < 0.001, $${\eta }_{p}^{2}$$ η p 2 = 0.31). Mainly the performance of the ball-on-bar task (p < 0.05, $${\eta }_{p}^{2}$$ η p 2 = 0.18–0.51) was modulated by MACS-levels, showing that children with MACS-level lll had worse bimanual coordination compared to children with MACS-level l and/or II. Ball-on-bar outcomes were highly related to task execution (r = − 0.75–0.70), whereas more interlimb differences of the object-hit task were moderately associated with a worse performance of the non-dominant hand (r = − 0.69–(− 0.53)). Conclusion This study gained first insight in important robotic tasks and outcome measures to quantify bimanual coordination deficits in children with uCP. The ball-on-bar task showed the most discriminative ability for both bimanual coupling and interlimb differences, while the object-hit and circuit tasks are unique to interlimb differences and bimanual coupling, respectively.