Scaling correction in pediatric center-of-pressure analysis: a dual-track computational framework for standardized balance assessment.

Objective: In this study, we developed and validated a standardized computational framework for pediatric balance assessment by integrating anthropometric normalization and a dual-track analytical architecture into a unified center-of-pressure (COP) signal-processing pipeline. Methods: This study included 90 typically developing children. They were stratified into age groups of 4, 6, and 10 years (n = 30 per group), balanced by sex. The children performed bipedal stance tasks under eyes-open and eyes-closed conditions. Static balance was assessed using a Zebris force platform (120 Hz). COP sway area, mediolateral displacement, and anteroposterior displacement were measured. Two parallel datasets were processed that included raw COP signals and leg length-normalized COP signals. Both datasets were subjected to 2 (sex) × 3 (age) analysis of variance to evaluate scaling sensitivity. Coefficients of variation were computed to assess system-level dispersion. Results: Anthropometric normalization markedly increased parameter sensitivity. Under the eyes-open condition, normalized COP sway area and anteroposterior displacement revealed age effects not detectable in the raw dataset, indicating scaling bias in uncorrected COP signals. The dual-track comparison confirmed that normalization functions as a computational validation component rather than a statistical add-on. For most COP parameters, the coefficients of variation exceeded 50%, indicating substantial interindividual variability consistent with developmental heterogeneity rather than measurement error. Conclusion: The proposed framework advances pediatric balance assessment from a descriptive developmental comparison to a standardized, bias-aware signal-processing architecture. By integrating anthropometric scaling correction and dual-track validation into a unified COP signal-processing pipeline, the framework enhances scaling sensitivity, inferential validity, and reproducibility. It is readily deployable within intelligent clinical screening systems and supports the development of age-stratified reference databases and early detection algorithms for pediatric neuromotor assessment.