Shoulder Joint

Participants watched a horizontal screen and held the handle of a robot that was placed underneath the screen, while a cloth prevented the sight of their arm (Fig 1A). The position of the robot-handle was registered with a sampling frequency of 200 Hz and a resolution of 0.3×10^-3 degrees on each joint of the shoulder which translates into a resolution in Cartesian coordinates of less than 0.2 mm. The registered signal was used to display a green cursor (diameter 6 mm) representing the handle position onto the screen with the help of a projector. Furthermore, a black origin and one of three possible red targets were alternately projected. The origin was positioned approximately 45 cm away from the eyes of the participant and the targets were positioned 10 cm away from the origin, either straight ahead or 45° to the left or the right. The origin as well as the targets had a diameter of 14 mm and the cursor one of 6 mm. The subjects were instructed to move the cursor accurately and quickly from the origin to the target and back. To control for the speed of the movements target color changed to green for actual trail times of 850ms ± 100ms and turned to blue or yellow when movements were too slow (>950ms) or too fast (>750ms), respectively. Intertrial intervals lasted for 1500ms.
After familiarisation with veridical and baseline without visual feedback, i.e. no cursor visible, all participants conducted six sets, each containing a baseline/washout block with veridical visual feedback, an adaptation block with rotated visual feedback (20°CW, 40°CW or 60°CW) and an inclusion and exclusion block without feedback. During each adaptation block, six clamp trials were inserted to test for the progression of recalibration (trial number 6, 19, 30, 39, 47 and 58). In those trials a perfect movement of the cursor from the starting to the target dot was displayed independent of the subjects´ movement. Each participant performed two consecutive sets for each rotation size with alternating order of inclusion and exclusion blocks. Before inclusion subjects were instructed to ‘use what was learned during adaptation’ and before exclusion subjects were asked to ‘refrain from using what was learned, perform movements as during baseline’. This order as well as rotation size order was randomised between participants. Between the third and fourth set there was a rest break of 5 min. Table 1 shows an overview of the experimental protocol.
After completion of the experiment all participants filled out a questionnaire as in Benson et al. (2011). Those participants who characterized the perturbation as a rotation or reported the use of a rotational strategy were considered to be explicitly aware of the distortion.

All patients underwent the shoulder arthroscopic surgery with the same surgical technique and all surgeries were performed by the same surgeons. After receiving general anesthesia combined with interscalene plexus block, all patients were placed in the lateral decubitus position. Normal saline containing epinephrine (1 mg/3 L) was used for intra-articular irrigation. A standardized arthroscopic approach for surgery was used with all patients.
First, an arthroscopic intra-articular examination was performed in all cases to probe whether the patients had other intra-articular lesions. We used 1 to 2 suture anchors (4.5 mm, TWINFIX Ultra PK Suture Anchor, Smith & Nephew) to repair the subscapularis if the subscapularis tendon was torn and recored the type of subscapularis tendon tears. Meanwhile, we performed a biceps tenotomy if the patients had a superior labrum anterior and posterior (SLAP) lesion or degenerative biceps tendon.
Then, via subacromial space arthroscopic vision, we performed acromiolpasty in cases with subacromial impingement syndorme and recored the type of postero-superior cuff tears. The double-row suture-bridge technique was used to repair the cuff tears (4.5 mm, Healix Healix Anchor System, Depuy and 4.5 mm, TWINFIX Ultra PK Suture Anchor, Smith & Nephew). Irreparable type C4 postero-superior cuff tears were eliminated from this study.
At the end of the operation, using a 14G puncture needle, 10 ml of TXA (100 mg/ml) or normal saline was injected into the shoulder joint through a posterior approach.

Data recording and processing were carried out using the software NI™ DIAdem (National Instruments, Austin, TX, United States). To prepare the data for evaluation, ACC signals were converted from volts to angles. All raw signals (force, pressure, ACC) were filtered with low pass Butterworth filter (filter order: 10, cutoff frequency for force and pressure: 3 Hz, for ACC: 1 Hz). Thereupon, the following force parameters were extracted. It is to be noted that the force was recorded in V and was transformed after extraction into torque (Nm) by using the formula M = F *r, where F is the force in N (converted by 1 V = 19.886 kg * 9.81 = 195.082 N) and r is the length of the individual rotational axis (lever) in m.

1) Maximal voluntary isometric contraction (MVIC)

The peak values of the MVIC trials were extracted. The highest values of the three MVIC trials before and of the two MVIC trials after the AF measurements were chosen as maxMVICpre (Nm) and maxMVICpost (Nm), respectively, and were used for further consideration.

2) Parameters of Adaptive Force

Exemplary force and angle signals of the arm and lever for one AF measurement are shown in Figure 2, illustrating the main aspects of the evaluation of AF parameters.

2.1) Maximal Adaptive Force (AF_max)

The peak value of each AF trial is referred to as AF_max (Nm).

2.2) Maximal isometric Adaptive Force (AFiso_max)

AFiso_max (Nm) defines the force value at the moment of first yielding of the forearm (breaking point). To determine this, a standardized algorithm was used according to Dech et al. (2021) (link). The main criterion for AFiso_max was that a holding isometric action was present from the beginning of the measurement. Thus, the necessary defined conditions were yielding of the forearm ≤2° (isometric action is still acceptable) and a push back of lever I ≤ 0.3° (pushing isometric action or concentric muscular action were excluded thereby, which was not the case in the present study). The limit values have been set in previous investigations (Dech et al., 2021 (link)). To determine the exact breaking point, the angles of the arm and lever were used. The second derivation was calculated from these to find the point of greatest curvature. AFiso_max was defined as the highest force value between the last maximum in angle signals (arm or lever) and the point of the subsequent greatest curvature before the forearm yielded more than 2° (arm angle). The deviation of the forearm (arm angle) at the beginning as shown in Figure 2 occurred regularly as soon as the force increased. It was presumably due to the cushion of the interface, the participant’s hand, elbow, or shoulder joint. Lever I did not show this behavior; on the contrary, its angle showed a yielding mostly always from the beginning. However, it was decisive that the forearm was in an isometric position. Due to the still-novel algorithm, the determined AFiso_max values were also checked visually. In 359 of 360 trials, the detected AFiso_max corresponds to the visual assessment of the breaking point.
Different ratios were calculated for further consideration to gather information on the relation of torque parameters: $\frac{{\mathrm{A}\mathrm{F}\mathrm{i}\mathrm{s}\mathrm{o}}_{\max }}{{\mathrm{A}\mathrm{F}}_{\max }}$ (%), $\frac{{\mathrm{A}\mathrm{F}\mathrm{i}\mathrm{s}\mathrm{o}}_{\max }}{\mathrm{m}\mathrm{a}\mathrm{x}\mathrm{M}\mathrm{V}\mathrm{I}\mathrm{C}\mathrm{p}\mathrm{r}\mathrm{e}}$ (%), and $\frac{{\mathrm{A}\mathrm{F}}_{\max }}{\mathrm{m}\mathrm{a}\mathrm{x}\mathrm{M}\mathrm{V}\mathrm{I}\mathrm{C}\mathrm{p}\mathrm{r}\mathrm{e}}$ (%).