A predictive model for classifying low back pain status based on lumbopelvic kinematics measured using inertial measurement units: a cross-sectional study

Abstract

Background: Low back pain (LBP) is a leading cause of disability worldwide. Impaired lumbopelvic control contributes to chronic and recurrent LBP, often presenting as aberrant movement patterns. This study aimed to investigate whether inertial measurement units (IMUs) can classify individuals with no LBP (NoLBP), chronic LBP (CLBP), and a history of LBP (HxLBP) based on lumbopelvic kinematics. Methods: A total of 141 participants (47 per group) performed ten standardized lumbopelvic movement control tests while wearing IMU sensors. Kinematic parameters, including mean velocity (MV), peak-to-peak amplitude (P2P), and area under the curve (AUC) of acceleration, were extracted. One-way ANOVA was used to compare kinematic differences across groups, and binary logistic regression models were developed to identify predictors for classification. Robustness analyses using 10-fold cross-validation with the least absolute shrinkage and selection operator (LASSO) were also performed. Results: Significant group differences were found in MV, P2P, and AUC across multiple movement tests (P < 0.05). The most pronounced differences were observed between NoLBP and CLBP: individuals with CLBP were characterized by slower trunk flexion (odds ratio [OR] = 0.94, 95% CI: 0.90–0.98), greater AUC during prone hip rotation (OR = 2.78, 95% CI: 1.45–5.34), and greater P2P during trunk rotation (OR = 1.32, 95% CI: 1.12–1.55). Robustness analyses confirmed the robustness and stability of the classification models. Conclusion: IMU-derived kinematic parameters provide objective measures of impaired movement control and may support clinical identification of individuals at risk for chronic or recurrent LBP.