Sex related gait dynamics are moderated by age

One of the earliest studies to compare men's and women's walking patterns was conducted on a New York City street in 1961. Sixty-four years and thousands of studies later, inertial measurement units, among other technologies, have now become commonplace tools for capturing gait differences between men and women across young, middle-aged, and older adults.

We focused our attention on a set of so-called nonlinear time series analysis metrics including the Hurst exponent, sample entropy, and recurrence quantification analysis. These analyses were applied to time series of stride intervals, stride lengths, and ranges of motion from the hip, knee, and ankle, as well as to segment angles of the thigh, shank, and foot, derived from 2199 walking trials containing at least 174 strides each, recorded at 200 Hz.

Subtle main effects on the gait dynamics of men and women were discovered, depending on the nonlinear gait metric and analyzed time series. None of the interactions between age and sex were present. There were also subtle effects of subsequent trials and data collection sessions that were dependent on the nonlinear gait metric and analyzed time series.

Implications of these small, natural differences between age, sex, session, and trial number are discussed. Speculations about their potential impact, limitations for certain algorithms, and a general call for the development of a collaborative database of nonlinear time series analysis metrics for gait, are also provided.