Regression models that predict anterior-posterior moment at the boot top (M_y^b), anterior-posterior force on the tibia (F_x^kb), varus-valgus moment at the knee (M_x^k), and torsional moment along the tibia (M_z^kb) are developed using data from slalom skiing runs for a 15-subject sample. The regressor variables are the lateral and vertical force components measured at the toe and heel bindings. The M_y^b and M_x^k regression model coefficients, standard deviations, and correlation coefficients (r) are dependent on the distance from the center of ankle flexion to the boot top (l_b) and to the knee (l_k).

It is recommended that the vertical and lateral force components at the toe and heel bindings be considered in future binding designs, because (i) these four force components are the smallest set of forces measured at the bindings required to predict M_y^b, F_x^kb, M_x^k, and M_z^kb with correlation coefficient r > 0.9 for this subject sample, and (ii) measurements of additional force or moment components do not necessarily increase model accuracy enough to warrant the increased model complexity.

Larger l_b (with l_b < 0.4 m) likely increases the capability of the state-of-the-art heel binding to control M_y^b for this subject sample, because the standard deviation decreases as l_b increases for the regression model.

The regression models are applied to skiing runs during which slalom turning, forward/backward leaning, falling, and jumping maneuvers are executed by the subjects. The average mean residuals and average standard deviations of residuals (s) are shown graphically as measures of regression model accuracy over these different skiing motions. The average standard deviation of residual of the regression models can vary by more than 100% when the models are applied to different skiing motions. Thus, caution must be exercised if the models are extrapolated to different skiing conditions where no verification of model accuracy exists.