AI-Enabled Telerehabilitation for Chronic Nonspecific Neck Pain: Multicenter Randomized Controlled Trial.

Background: Exercise therapy is effective in managing chronic nonspecific neck pain (CNP). Nevertheless, the high burden of CNP continues to place pressure on existing rehabilitation services, especially in remote or resource-constrained settings where continuous and standardized physical therapy may be difficult to access. By combining home-based exercise with remote supervision and movement feedback, AI-enabled telerehabilitation may broaden access to rehabilitation and support the continued management of CNP. Objective: This study examined whether an AI-enabled telerehabilitation program was noninferior to standard outpatient rehabilitation for improving neck-related disability in individuals with CNP and compared the effects of the 2 interventions on pain, health-related quality of life, and cervical range of motion. Methods: This multicenter, prospective, parallel-group randomized noninferiority trial was conducted at Peking Union Medical College Hospital and Beijing Jishuitan Hospital. Patients with CNP of at least 3 months' duration were randomized 1:1 to a telerehabilitation group (TG) or outpatient rehabilitation group (OG) using concealed, web-based, variable blocks stratified by center and baseline Neck Disability Index (NDI) severity. Both groups completed identical 24-week strengthening and stretching programs (3 sessions per wk; ≈30 min per session) and were assessed at weeks 0, 4, 8, 12, and 24. TG trained at home using sensor-based movement recognition and threshold rules; OG received face-to-face outpatient rehabilitation. Week-24 NDI was the primary outcome; Numeric Pain Rating Scale (NPRS), EQ-5D, and cervical range of motion were secondary. Noninferiority required the upper 95% CI bound of the TG-OG NDI difference to be <10 percentage points (5/50). Analyses were intention-to-treat; assessors and statisticians were blinded. Results: Of 312 individuals screened, 116 were randomized (58 per group). At week 24, 55 TG and 53 OG participants completed assessments; all 116 participants were included in the primary intention-to-treat analysis using observed or imputed data. The between-group differences in NDI (TG - OG) were -2.8 (95% CI -6.3 to 0.7; Conclusions: Over 24 weeks, AI-enabled telerehabilitation was noninferior to exercise-matched outpatient rehabilitation for neck-related disability, with broadly comparable improvements in pain, health-related quality of life, and cervical range of motion. AI-enabled telerehabilitation may be a practical alternative for patients with CNP who are unable to attend outpatient rehabilitation.