JMIR Publications

ABSTRACT

Background:

CNP means chronic non-specific neck pain, imposing a substantial societal burden. Conventional outpatient rehabilitation is constrained by limited access and continuity, particularly for patients distant from clinics or facing transportation barriers. Smartphone-based rehabilitation integrating sensors and remote therapist oversight may offer a scalable means to enhance access and exercise adherence while preserving clinical efficacy.

Objective:

The objective is to compare the clinical outcomes of an AI-enabled tele-rehab program with those of outpatient rehabilitation in individuals with CNP.

Methods:

In this multicenter, randomized controlled trial, participants were randomly assigned 1:1 to either the tele-rehab group (TG) or the outpatient-rehab group (OG). Both underwent standardized strengthening and stretching three times weekly (~30 min/session) over the intervention, with assessments at baseline and scheduled follow-ups. TG performed exercises and completed questionnaires at home via a remote platform with sensor-based feedback and therapist adjustments; OG received equivalent in-clinic instruction. The Neck Disability Index (NDI) served as the primary endpoint. Secondary evaluations included the Numeric Pain Rating Scale (NPRS), EQ-5D for health-related quality of life, and cervical range of motion. Non-inferiority analysis was conducted based on the differences observed between the two groups, and statistical significance was determined P-value < 0.05.Outcome assessors and statisticians were blinded.

Results:

Of 312 screened, 116 were randomized to telerehabilitation-based exercise (TG; n=58) or outpatient-based exercise (OG; n=58). Baseline characteristics were comparable with P>.05. Both groups exhibited progressive, clinically meaningful NDI reductions for all follow up time points with P>0.1; between-group differences (TG−OG) were −2.8 by the 12th week (95% CI −6.3 to 0.7; P=.11) and −2.5 by the 24th week (95% CI −6.2 to 1.2; P=.15), supporting noninferiority of TG. NPRS improved similarly with between-group differences of −0.1 by the 12th week (95% CI −0.5 to 0.3; P=.64) and −0.3 by the 24th week (95% CI −0.7 to 0.1; P=.16). EQ-5D improvements were comparable: +0.02 by the 12th week (95% CI −0.04 to 0.08; P=.50) and +0.01 by the 24th week (95% CI −0.05 to 0.07; P=.73). Cervical ROM exhibited time-dependent gains in both arms; between-group contrasts were neither statistically nor clinically significant at the 12- and 24-week assessments.

Conclusions:

Compared with standard outpatient rehabilitation, AI-enabled telerehabilitation demonstrated non-inferior clinical outcomes in patients with CNP and offering better scalability. These findings support telerehabilitation as a viable alternative for long-term management of CNP across clinical and home settings. Clinical Trial: Chinese Clinical Trial Registry ChiCTR2500112693; https://www.chictr.org.cn/showproj.html?proj=286440