Multimodal Exercise Therapy for Non-Surgical Intervention of Nonspecific Low Back Pain.

Low Back Pain, Machine Learning, Non-surgical Treatment, Paraspinal Muscles, Radiomics

Nonsteroidal anti-inflammatory drugs, Multidimensional exercise intervention

This multicenter, assessor-blinded, two-arm parallel randomized controlled trial (N = 314) will compare the efficacy and safety of a 6-week multidimensional exercise program plus usual pharmacological care (experimental arm) versus usual pharmacological care alone (control arm) in adults ≥ 60 years with chronic non-specific low-back pain (LBP) and imaging evidence of paraspinal muscle degeneration. The primary endpoint is change in Oswestry Disability Index (ODI) at 12 months. Secondary endpoints include pain VAS, JOA score, recurrence rate, and patient satisfaction measured repeatedly to 12 months. Advanced MRI radiomics and machine-learning algorithms will be used to build a "paraspinal muscle imaging-function-prognosis" prediction model and an open-access web tool for risk stratification. The study will generate a standardized, evidence-based non-operative care pathway for chronic LBP driven by paraspinal muscle degeneration

Primary: to determine whether a 6-week supervised home-based multidimensional exercise program (aerobic, strength, balance, flexibility) added to short-term NSAIDs improves 12-month functional outcome (ODI) compared with NSAIDs alone. Secondary: (1) compare pain intensity, quality of life, recurrence, and safety between arms; (2) develop and validate a radiomics-plus-machine-learning model that uses baseline MRI features of paraspinal muscles, demographic, clinical, and functional variables to predict treatment response and recurrence risk; (3) integrate the validated model with the exercise intervention to create a risk-stratified, early-intervention clinical pathway for chronic LBP. Study Design：Multicenter, prospective, assessor-blinded, superiority RCT with 1:1 allocation, conducted at 4 tertiary hospitals and 1 academic research center in Beijing. Randomization (block sizes 4 and 6, stratified by center and sex) generated centrally via web-based system; allocation concealed from outcome assessors and statisticians. Population Inclusion: age ≥ 60 y; LBP ≥ 3 months; ODI 20-60; MRI evidence of paraspinal muscle fatty infiltration ≥ Goutallier grade 2; no surgical indication; able to exercise. Exclusion: specific spinal pathology (infection, tumor, fracture, severe deformity); previous lumbar surgery; severe comorbidities limiting exercise; cognitive impairment; current participation in another trial. Interventions Experimental: 6-week multidimensional exercise (5 days/week, 30-40 min) plus 4-week oral NSAIDs (celecoxib or acetaminophen, dose-adjusted). Exercise comprises (1) low-intensity aerobic walking; (2) light-load high-repetition strength training (upper limb, lower limb, core); (3) balance and proprioceptive drills; (4) stretching. Printed manual + training diary; weekly phone/video supervision; compliance algorithm (≥ 75 % overall score required). Control: same NSAID regimen alone. Both groups receive standardized education booklet on posture and activity pacing. Outcomes Primary: ODI at 12 months (minimal clinically important difference 10 points). Secondary: VAS pain, JOA score at 1, 3, 6, 12 months; recurrence (≥ 24 h LBP with VAS \> 2) within 12 months; patient satisfaction (NASS questionnaire); adverse events; compliance. Imaging: axial MRI (3 T) at L1-5; semi-automated segmentation (3D-Slicer) to quantify cross-sectional area, fat fraction, radiomic features (shape, first-order, GLCM, GLRLM, GLSZM, GLDM). Functional: Biering-Sørensen endurance test; surface electromyography of multifidus. Biomarkers: baseline bloods (CBC, renal, hepatic panels). Sample Size 314 participants (157 per arm) provide 80 % power (α = 0.025 one-sided) to detect 15-point ODI difference (SD 15), allowing 10 % attrition. Data Analysis Intention-to-treat primary analysis: ANCOVA adjusted for baseline ODI, sex, center. Secondary longitudinal scores analyzed with mixed-effects models; recurrence with Kaplan-Meier and Cox regression.Radiomics: 7:3 split; 100× bootstrap; LASSO feature selection; multivariable modeling (logistic regression, random forest, XGBoost, LightGBM, MLP); performance reported as AUC, sensitivity, specificity, PPV, NPV, F1; SHAP/LIME for interpretability. Internal 5-fold cross-validation; final model deployed as web calculator. Quality & Ethics Protocol approved by central ethics committee; registered at Chictr.org.cn. Written informed consent. Data entry via validated EpiData system; double verification; 5-year retention. Adverse events monitored by independent DSMB. Study conducted in accordance with Helsinki Declaration and ICH-GCP.

No related papers found yet.

China