*2.1. Patient Recruitment*

Inclusion criteria were age ≥ 50 years, displaced or unstable three- or four-part fracture of the proximal humerus (except isolated displaced fractures of the greater or lesser tuberosity) treated with a plate and screw osteosynthesis (PHILOS locking plate—with or without screw augmentation) within 10 days after injury, ability to understand the content of the study and the patient consent form and voluntary signed informed consent.

Exclusion criteria were previous proximal humerus fracture on the ipsilateral limb, humeral head impression/splitting fracture, fibula grafting, bone block or any other noncement augmentation of the PHILOS locking plate fixation, associated nerve or vessel injury, serious fracture fixation issues such as too long screw, screw perforation through the humeral head, or a broken screw or implant recognized directly on the first postoperative X-ray. Other exclusion criteria were severe systematic diseases rated in class 4 and higher of the American Society of Anesthesiologists (ASA) physical status classification, substance abuse, prisoner, participation in another medical device or product study in the past month that could affect this study, pregnancy, or pacemaker.

Patient data including age, gender, height, weight, residential status, injury side, arm dominance and fracture type were collected at recruitment.

#### *2.2. Postoperative Protocol*

The two university hospitals used different postoperative rehabilitation protocols according to their standard of care. In hospital H1, the patients were treated with a sling for 3 weeks and were only allowed passive and active-assisted mobilization under supervision of a physiotherapist for the first 3 weeks. Patients in hospital H2 were treated without a sling and allowed to mobilize without restrictions immediately. Physiotherapy was started immediately postoperatively and prescribed 2–3 times per week in both hospitals.

#### *2.3. Activity Tracking Apparatus and Procedure*

Accelerometer sensors (AX3, Axivity Ltd., Newcastle upon Tyne, UK) [23] (Figure 1, left) were used to measure shoulder activity continuously (24/7) for 6 weeks after the operation in two consecutive 3-week periods. The length of the measurement period was determined by the sensor's battery and memory capacities. The first period started at the latest 4 days postoperatively and ended at 21 ± 3 days, the second period started at the same visit and ended at 42 ± 3 days. Two sensors were used for each patient and period. One sensor was attached to the upper arm of the treated side, and another was located at

the chest and served as a reference (Figure 1, right), allowing evaluation of the shoulder angle as the orientation difference between both devices. Data recording was performed at 50 Hz frequency within ±4 g limits that were deemed suitable in a pilot evaluation. The sensors were attached to the skin using a dedicated certified medical-grade adhesive tape (3M 4077, 3M Medical Materials & Technologies, Oakdale, CA, USA). Attachment (directly postoperatively and at the 3-week follow-up visit) and detachment (at the 3-week and 6-week follow-up visits) were performed by trained study personnel according to standard operating procedures to ensure consistent sensor location and alignment. The start and end time points of a given period were marked by knocking five times synchronously at both arm and chest sensors. The patients were allowed to follow their normal daily activities including showering with the attached device. At the end of the measurement periods and after detachment, sensor data were downloaded using the Open Movement GUI software (Open Movement project).

'

" " " " – – **Figure 1.** Accelerometer sensors (Axivity AX3, (**left**), source: https://axivity.com/) were attached to the patients (illustrated on the (**right**)), on the chest ("1") and on the back of the upper arm on the treated side ("2"). Each sensor had its own coordinate system, indicated with the red–green– blue arrows.
