Bench-stepping training improves stair-walking dynamics in older women: evidence from an exploratory nonlinear kinematic analysis

Introduction: Stair walking challenges balance and coordination in older people. Bench-stepping training improves stair climbing speed in healthy older women. This study assessed whether bench-stepping also improves dynamic balance and movement complexity during stair walking. Methods: Stair walking data were obtained from a previous study involving 45 healthy older women (69y+/-4) that assessed the effects of a 12-week bench-stepping intervention with non-training controls. Centre-of-mass acceleration was measured during stair ascent and descent. Linear dynamics included time, acceleration magnitude, and harmonic ratios (HR; indicating symmetry). Movement complexity was quantified using nonlinear dynamics including sample entropy (SE), recurrence quantification analysis (RQA), and fractal dimension (FD). Results: For stair ascent, increased speed (p =0.018, R2partial =0.093,) was accompanied by proportional increases in acceleration magnitudes (p=<0.039, R2partial =0.078-0.101). SE decreased more in the intervention group (p=<0.012, R2partial =0.049-0.101), indicating more predictable dynamics. In contrast, for stair descent, no changes in speed or acceleration magnitudes were observed. However, SE (p =0.001, R2partial =0.082) and maximum RQA line length (p= 0.008, R2partial =0.057) of vertical acceleration increased significantly compared to controls, indicating lower predictability and more persistent recurring patterns. No significant changes were found for other outcomes. Exploratory factor analysis revealed distinct differences in motor behaviour between stair ascent and descent. Conclusion: Bench-stepping training induced measurable changes in stair walking dynamics. Specifically, sample entropy shows potential as a sensitive marker of altered motor complexity, particularly of vertical accelerations. Interestingly, the direction of changes in unpredictability differed between stair ascent and descent, suggesting different underlying control strategies.