Problem	The adoption of ML models into clinical workflows is lacking because traditional ML evaluation metrics fail to accurately assess how useful a model will be in practice.
What is Already Known	Prior work has simulated individual model impact in the context of specific care delivery workflows. However, these efforts have limited generalizability to other models/workflows and exhibit overreliance on non-modifiable assumptions.
What This Paper Adds	Our contribution builds on prior work through the development of a flexible, reusable set of methods that allow for the systematic quantification of the usefulness of ML models by simulating their corresponding care management workflows. The APLUS library can help hospitals to better evaluate which models are worthy of deployment and identify the best strategies for integrating such models into clinical workflows.