Intelligent sampling of high‐dimensional joint mobility space for analysis of articular function

Abstract

Abstract Studies of joint structure and function have played a central role in understanding palaeobiology and major functional transitions in evolutionary history, where fossilized hard parts (e.g. bones) are often all that remain. Current digital methods for quantifying articular function depend on exhaustively sampling all possible joint poses, but this is computationally intensive, becoming prohibitively so with more degrees of freedom. This has impeded more sophisticated analyses or broader application of these methods to more diverse questions and species. The present study introduces ‘APSE’ (Accelerated Pose Searching with Electrostatics), an iterative algorithm for rapidly and intelligently sampling high‐dimensional joint pose space to quantify articular function and mobility. Key features of the algorithm are: known viable joint poses inform the search for new poses in successive generations; the search preferentially targets under‐explored regions of pose space, while avoiding over‐explored regions; large swaths of inviable pose space are never evaluated, thus minimizing wasted time; and parallelizability. The efficacy of the algorithm was benchmarked with diverse theoretical and biological joints. As a case study to demonstrate its utility, APSE was used to investigate mobility in the enigmatic shoulder joint of the extinct ‘pelycosaur’‐grade synapsid Dimetrodon, the function of which has been widely debated. APSE produces high‐dimensional joint mobility assessments with greater precision than state‐of‐the‐art exhaustive sampling methods. More viable poses are identified at a greater sampling density, but in a fraction of the time taken by current methods (hours, not days or weeks). Results for Dimetrodon shoulder mobility indicate strong coupling between most degrees of freedom across the joint's full range of motion, a stark contrast to the flexible shoulder of most extant tetrapods. APSE provides a time‐efficient means to quantitatively measure articular function and mobility, especially when more degrees of freedom are considered. By greatly reducing computational requirements, APSE lowers the barrier to researchers seeking to undertake more complex or more numerous analyses of articular function in modern and extinct animals. This will accelerate the pace of research in comparative or evolutionary analyses of joint and animal function.