To judge feasibility and acceptability, the study was designed as parallel-group design with a group of CP patients and an age-matched group of able-bodied counterparts. To asses reproducibility, two visits were planned on separate days for each participant. The study setting was the UniReha’s Queen Rania Rehabilitation Center in Cologne. Prior to study commencement, it had been approved by the ethics committee of the university hospital cologne and registered with the German Register for Clinical Studies (DRKS) with the index DRKS00031107 (www.drks.de/DRKS00031107).

All participants gave informed, written consent before participation. In case of underaged participants, informed and written consent was obtained from their parents. The patient group consisted of twelve adolescents between 8 and 18 years with diagnosed unilateral or bilateral spastic cerebral palsy (CP). Participation was offered to the adolescents with CP upon recommendation of the treating physical therapists based on the patient’s cognitive and motor skills. Exclusion criterion was reduced knee and ankle joint mobility inhibiting the measurement position of 90∘ knee flexion and neutral ankle position. For each patient healthy participants with matching sex and age (±1 year) were recruited for the control group.

The custom-built measurement device consisted of a adjustable seat and a fixed dynamometer (Figure 1). The dynamometer’s working principle was based on a foot plate that was rotatable around a fixed axis of rotation (AOR) and rested on a force sensor under the front foot. The participants’ medial malleolus was aligned with the dynamometer’s AOR. To achieve the anticipated measurement position, the seat platform’s height above the AOR was adjusted to the length of the shank (distance from medial malleolus to medial knee joint cleft). The distance from the seat to the dynamometer was adjusted until the shank was oriented vertically (neutral ankle joint angle). The heel was prevented from lifting during force generation by stabilizing the distal thigh from the ventral side by a padded metal bar, serving as resistance (Figure 1). A soft foam pad in front of the foot served as landmark for the foot position, in case the foot has been moved between trials. The foot has not been fixated further to allow motion in case of arising spasms. To increase acceptability and compliance of the participants the whole measurement was gamified. The dynamometer part of the measurment device was covered in a rocket-style aluminium shell and all tasks were visualized in a torque controlled video game in which a digital rocket rose up with increasing plantarflexion torque.

There were three tasks given with three trials per task, each of them demanding different types of isometric plantarflexion contractions. In participants with unilateral CP, the affected leg was chosen, while the leg with better movement was chosen in patients with bilateral CP. The control group participants’ right leg was examined. The first task was the maximum voluntary isometric contraction (MVC). Participants were asked to drive the game rocket as high as possible, meaning to physically exert their largest possible torque. During trial 2 and 3, the maximum previously achieved torque was always displayed. From the overall maximum achieved, submaximal levels were defined at 20%, 40%, and 60% MVC. In three different trials these levels were presented for 16 s as horizontal stars in the second task “step contraction.” To accomplish this task, participants were asked to collect 30 stars, that were arranged in a step-wise fashion, with five stars at rest, followed by 20 stars at the desired contraction level, back to five stars at rest. During the third and last task “ramp contractions” participants performed continuously increasing and decreasing isometric contractions for eight seconds each from rest to 100% MVC and back. This task was visualized by stars arranged in a triangular shape (Figure 1).

The repeatability measurements were taken after an interval of at least 24 h.

At the beginning of the first visit height and weight data, as well as the Achilles tendon lever arm (ATLA) length were collected. Afterwards the participants have been equipped with electromyography (EMG) electrodes at the soleus, lateral gastrocnemius and tibialis anterior muscle according to SENIAM standards (www.seniam.org) and with a goniometer (Noraxon U.S.A. Inc., Scottdale, United states) at the ankle joint. Force, EMG, and goniometer signals as well as a trigger signal of a custom-built hand trigger have been transmitted to a Noraxon system (Noraxon U.S.A. Inc., Scottdale, United states) and recorded during the whole measurement using the software myoResearch®version 1.08 Master Edition and a sampling rate of 1,000 Hz. The hand trigger was pressed at the beginning and end of each trial.

All data processing has been performed using the programming language R version 4.3.2 (www.r-project.org) with the RStudio development environment (14). Force and joint angle signals were filtered using a fourth order Butterworth low-pass filter with a cutoff frequency of 5 Hz. The force signal was converted to Nm via calibration with a 1 m lever and a 10 kg mass. Afterwards the data files of each subject cut into individual trial recordings with the hand trigger signals. Resting torque was defined in each of the three MVC trials as mean torque of a manually defining 200 ms period before contraction with no visible contraction in the EMG signals. The MVC torque of each trial was calculated as average torque of the torque plateau. The plateau was defined as the area where the torque reached over 90% of the maximal torque. The maximal MVC value out of the three trials was then set as total MVC to calculate the contraction levels for further trials of tasks 2 and 3. For all tasks performance parameters have been calculated and statistically analyzed. In case of comparable performances between groups, additional muscle parameters about fascicle mechanics and muscle activation have been analyzed.

Statistical analyses followed the “per-protocol” principle, i.e., incomplete data sets were discarded. Repeatability of the MVC and the performance parameters steadiness and slope has been tested using the two-way agreement-type intraclass correlation coefficient (ICC). Between-group comparison within the visits has been performed using Mann-Whitney U-test with Holm correction for multiple comparisons. The statistical significance level was set to 5%.

During the first visit, participants of the CP group reached a MVC of 1.0 g while the control group reached 4.3 g (p<0.001). Similar results have been found for visit two (1.4 g vs. 4.9 g, p<0.001). Plantarflexion torque fluctuated more in the CP group in all tasks at visit 1. During step contractions CP group’s torque fluctuated by 52.2% MVC vs. 6.1% MVC in the control group (p<0.001). Similar differences could be seen during ascending ramp contractions (26.3% MVC vs. 7.0% MVC, p=0.005) and descending ramp contractions (34.1% MVC vs. 8.1% MVC, p=0.025). At visit 2 group differences in steadiness decreased but remained significant for step contractions (20.6% MVC vs. 5.3% MVC, p<0.001) and descending ramp contractions (15.9% MVC vs. 7.7% MVC, p=0.025) but vanished for ascending ramp contractions (Figure 4). There were no significant differences between groups in slopes during all tasks and visits.

MVC showed excellent repeatability between the measurements with an ICC of 0.94 (p<0.001), whilst all performance parameters of all tasks showed poor ICCs (Table 2). Consistent with the previously mentioned better performance of the control group during visit 1 and 2, we see only little differences in fluctuations between visits in this group (Figure 5). By contrast, almost all participants of the CP group improve their performance during step and ramp contractions from visit 1 to visit 2 (Figures 5B–D). Despite the lack of any significant group differences between slopes in both visits, there is a visible trend in all tasks of the group to converge (Figure 4).