Keynote Dr. Aaron Likens

Abstract

Time series data collected in movement science are notoriously noisy. Spatiotemporal measures collected during locomotion or maintenance of upright posture often vary considerably so over several minutes of observation. Traditional linear statistics such as the mean and standard deviation often fail to capture these time varying properties. A key feature of biological signals such as heart rate, neural activity, and human walking is that the underlying processes are complex, requiring coordination across many timescales. These scales range from milliseconds typically relevant for neuroscience to the multiple minutes that make up a single bout of walking. Thus, analytical methods are required to address the multiscale nature of human movement and physiological data. In this keynote (and accompanying workshop), I will discuss the ongoing development of class of tools ideal for studying this complexity that includes well-worn methods such as multifractal analysis as well as recent tools that extend fractal regression methods to the multivariate time series.

Bio

Dr. Likens is a quantitatively trained movement scientist with broad education and training in biomechanics, cognitive science, statistics, and motor control. As Director of the Nonlinear Analysis Core (NONAN), he has extensive expertise in time series analysis and statistical modeling. A major focus of his research is to understand how statistical patterns in gait and posture change over natural course of aging and as the result of disease and injury. Other recent work has focused on developing new time series analysis methods for studying human movements and other forms of behavior. His work is funded by the NIH, National Strategic Research Institute (NSRI), and the National Science.