Considerations for the use of functional principal components analysis in sports biomechanics: examples from on-water rowing

Abstract

The proliferation of new biomechanical technology in laboratory and field settings facilitates the capture of data-sets consisting of complex time-series. An understanding of the appropriate statistical approaches for analysing and interpreting these data-sets is required and the functional data analysis (FDA) family of statistical techniques has emerged in the biomechanical literature. Given the use of FDA is currently in its infancy with biomechanical data, this paper will form the first of a two part series aiming to address practical issues surrounding the application of FDA techniques in biomechanics. This work focuses on functional principal components analysis (fPCA), which is explored using existing literature and sample data from an on-water rowing database. In particular methodological considerations for the implementation of fPCA such as temporal normalisation of data, removal of unwanted forms of variation in a data-set and documented methods for preserving the original temporal properties within a set of curves are explored in detail as a part of this review. Limitations and strengths of the technique are outlined and recommendations are provided to encourage the appropriate use of fPCA within the field of applied sports biomechanics.

A drop landing screening approach to monitor an individual using functional data analysis: An ACL injury case study

“…No temporal normalisation strategies were applied in an effort to retain the original temporal properties of the data set. 25 Each of these 0.5 s vGRF curves resulted in a total of 27 (six sessions x three landings) time-series, which were inputted into an fPCA.…”

Shape is only a weak predictor of deep knee flexion kinematics in healthy and osteoarthritic knees

“…Knee kinematics are complex and typically described as either time-series or angle-angle waveforms. 18,19 However, standard analysis techniques for describing and comparing knee kinematics typically focus on predefined features such as maxima, minima, specific values, and slopes 19,20 . The use of discrete data points neglects the entire waveform.…”

Call to increase statistical collaboration in sports science, sport and exercise medicine and sports physiotherapy

“…In another example, 26 modifications to the application of principal components analysis (PCA) applied within functional data analysis 40 were proposed for the context of exploring high-dimensional kinematic sports science data. [41][42][43][44] PCA estimates the principal components of a set of curves whose measured values are stored in a data matrix such that each row holds the data for an individual curve. In a recent preprint on SportRxiv, 26 the author argues that estimating the principal components of the data matrix, such that each column holds the measured values of an individual curve, is more appropriate.…”