Digital Therapeutics for Postoperative Rehabilitation after Arthroscopic Rotator Cuff Repair: A Single-Center,...

Home-based rehabilitation after arthroscopic rotator cuff repair (ARCR) often lacks objective monitoring. To address this limitation, we developed the ANAPA digital rehabilitation (DR) system, which includes a patient-facing mobile application for clinical efficacy evaluation (ANAPA ME) and a physician-facing measurement application for technical assessment (ANAPA PS). This study compared the safety and preliminary efficacy of ANAPA ME with conventional YouTube-based rehabilitation (CR), with an additional validation of the automated measurement functions in ANAPA PS.

This single-center, assessor-blinded, randomized pilot study included 18 patients who had undergone ARCR. The participants were allocated to the DR group (n = 9) or CR group (n = 9) and followed identical postoperative rotator cuff exercise protocols. The program was conducted for 24 weeks postoperatively. The shoulder range of motion (ROM), muscle strength, and clinical outcomes (visual analog scale pain score, Constant shoulder score, Shoulder Pain and Disability Index, and EuroQol-5 Dimension-5 Level questionnaire) were evaluated at baseline and at follow-up up to 24 weeks postoperatively. Ultrasonography was performed at 24 weeks to evaluate the integrity of the repaired tendons. Automated ROM and muscle strength measurements of ANAPA PS were validated against digital goniometry and hand-held dynamometry at the 6-month follow-up, and measurement consistency was assessed using intraclass correlation coefficients (ICCs) and coefficients of variation (CV).

Both the DR and CR groups demonstrated significant within-group improvements in ROM and clinical scores (p < 0.05). The DR group showed significantly greater improvements in specific ROM parameters, including forward flexion at 24 weeks and abduction, abduction-external rotation, and internal rotation at 18 and 24 weeks, compared with the CR group (p < 0.05). At 18 weeks, the internal rotation muscle strength was also significantly higher in the DR group than in the CR group (p < 0.05). A significant time-by-group interaction was observed for abduction, side-external rotation, and internal rotation (p < 0.05), suggesting a more consistent improvement pattern over time in the DR group compared with the CR group. In supplementary validation, ANAPA PS measurements showed high agreement with reference instruments (mean absolute error of 3.14° for ROM and 0.15-0.22 kgf for muscle strength). In consistency analysis, ANAPA PS showed higher repeatability and lower variability than digital goniometry (ICC = 0.99 vs. 0.97; CV = 1.61% vs. 11.7%, p < 0.001) and hand-held dynamometry (ICC = 0.97 vs. 0.39; CV = 2.1% vs. 13.8%, p < 0.001).

In this exploratory pilot study, the ANAPA DR system demonstrated favorable trends compared with CR in patients after ARCR, suggesting the feasibility and safety of artificial intelligence-based digital therapeutics in postoperative shoulder rehabilitation. The supplementary validation further demonstrated the technical reliability and measurement consistency of the ANAPA PS automated measurement functions for monitoring rehabilitation outcomes.