Disclosed are systems and methods for building and using a data platform to perform intelligent treatment selection to assist in clinical decision making. The present disclosure relates to a server configured to transform a subject record, which characterizes various non-numerical features of a subject, into a transformed representation that is consumable by machine-learning or artificial intelligence models. The output of the machine-learning or artificial-intelligence models can be evaluated to enable the intelligent treatment selection and other enhanced analyses, such as, for example, automatically detecting and defining subject populations, predicting treatment protocols with likely positive responsiveness, and identifying the similarity or dissimilarity between subject records along one or more dimensions.