Many of medical devices come equipped with a communication interface. Over the years, there has been interest in leveraging these interfaces to add computers to the loop to aid in decision making and automatic application of interventions. Such systems, which we call Closed-Loop Assistants (CLAs), are intended to help clinicians manage the cognitive load that can arise as the complexity of patient management increases. We present an open-source framework for examining CLA-patient interactions through software simulations of the CLA with in-silico patients to enable early testing and validation of proposed physiology management strategies. We show how this framework can be used to test different strategies across a small patient population. Considering a patient population is important because inter-patient variability is one of the critical factors that can hamper the ability of a medical cyber-physical system like the CLA to meet its goals. The ability to explore this variability early in the design process therefore helps us in increasing robustness of the system.