Determining Predictability of Human Gait Parameters

Abstract

Walking is a physiological phenomenon which is mathematically described as optimised solution of multiconstrain equation for human motion. This research is carried out with the goal of establishing a relationship between anthropometric data and gait variables. The study consisted of trials in which each participant is asked to walk the length of the instrumented walkway, using a motion capture and analysis system, the kinematic and kinetic parameters of each trial were calculated. The peak points obtained from the data curves were used to generate regression fits. Both univariate and multivariate fits were made to study the correlation of human anthropometric data to gait parameters.The KA2, AA1 and AA2 fits have enough low goodness of fit (barring AA1, which shows a 14.5% goodness of fit and p <0.05 with respect to walking speed). Of all joint angles, the hip joint angles HA1, HA2 and HA3 show a good fit with respect to the leg length data >body weight data >height data (in decreasing order of goodness of fit). The GRF parameters show the highest R2 (FZ0: 0.899) and lowest p-values in the linear fit against body weight data. Besides the GRF parameters also display a good fit against height and leg length parameters. These mathematical dependability connecting the behavior of gait parameters to the independent variables can be predicted much more reliably when considering the anthropometric data.
Normal human walking is simulated as the multi-segment human skeleton model which is mathemathicaly optimised by inverse dynamics techniques, for the quantitative comparison of the experimentally measured and numerically calculated data. The model reproduces the significant results. The study is further extended for determining asymmetry in the limbs by using mean symmetry indices technique and the formulation of the equation for characteristics points with the simple anthropometric descriptors (input): age, weight, height, leg length and walking speed. The data generated from the multivariate regression fits of the gait parameters against anthropometric data. It is interesting to note that the R2 values of for the GRF parameters decreased, as compared to the univariate regression fit against body weight. However, the significance levels of all data show an overall increase (p-values lying in the range of 1×10􀀀4 to 1×10􀀀39). Still, the KA1 values do not approach the standard p-significance level, thus confirming the results from the univariate studies discussed earlier. For the multivariate regression study, the R2 values show high goodness of fit values and high significance levels as compared to those in univariate studies. Thus it may be concluded that these multiple regression fits have a much better significance in predicting the parameters than a single univariate regression study. The determined numerical dependability has potential application in the rehabilitation process, determining design parameters for prosthetic designing and human motion study.