Statistical Design and EDA tools for the Clinical Treatment of Anterior Cruciate Ligament (ACL) Tears

Ponente(s): Thomas Santner

This talk will demonstrate how simulator experiments can be combined with cadaver experiments to model the tensions of knee ligaments in an uninjured, intact knee. The simulator experiments use a special kind of deterministic \multi-body" model. The cadaver experiments use a specified loading pattern. While the output of multi-body codes are deterministic, they are ordinarily biased. We demonstrate a method to calibrate the simulator output using Gaussian process models to predict the uncorrected simulator output and its bias, so as to better predict ligament strains in the intact knee. These values suggest how surgeons can properly tension a torn ACL. This talk is joint work with members of the Biomechanics and Biomaterials Departments of the Hospital for Special Surgery in New York City.