In this study, Duncan et al. explore a machine learning model that uses patient demographics and imaging data to guide clinical decisions between cemented and cementless total knee arthroplasties (TKA). The study gathered data from over 250 primary TKA patients, all operated on by the same surgeon, to both train and test the model. The machine learning model employed an artificial neural network to predict whether a cemented or cementless TKA would be recommended, depending on the context of the case. Along with clinical predictions, the model incorporated local interpretable model-agnostic explanations (LIME) to provide a reasoning behind its recommendations. Upon testing, the model achieved 86.4% sensitivity, 97.4% specificity, and 93.3% accuracy when identifying the surgeon's operative decisions. Aside from performance metrics, the study revealed that there was no single set of factors that could reliably predict implant type, highlighting the advantage of using a machine learning model to integrate multiple variables. Findings showed that the model’s LIME feature helped surgeons to better grasp the rationale for the model’s outputs to make more informed operative decisions.