Communications in Information and Systems Volume 15, Number 1, 35–46, 2015 Switched systems as models for dynamic clinical trials Y. Du and C. Martin Dedicated to Wing Wong on the occasion of his 60th birthday The goal of the paper is to make a case for the use of switched systems as a model for the simulation of dynamic clinical trials. Dynamic clinical trials are characterized by the use of different treatments based on the response of a patient to the current treat- ment. We show that switched systems are, in fact, a suitable model for these trials. Keywords and Phrases: dynamic clinical trials, switching sys- tems, stability. 1. Introduction The choice of a treatment level and its type is one of the most important, yet difficult decisions made by a researcher in a clinical study. Dynamic treatment regimes that provide treatment to patients only if and when the treatment is needed and adjust the level of treatment to the patient’s need become the most relevant for this purpose. These regimes are based on decision rules that input patient outcomes collected during treatment and output recommended changes in assignment of the type and level of treat- ment. To execute this correctly a researcher needs to consistently monitor a patient’s reaction to the treatments and adjust the treatment level and type accordingly. The idea behind dynamic clinical trials is based on the standard treat- ments of difficult diseases. Drug A is given to the patient and if the patient responds to the treatment then no further action is needed. However if the patient fails to respond in totality to drug A then the treatment will be switched to Drug B. In contrast to classical trials in which every subject is assigned to a fixed treatment, dynamic trials are designed to allow the treatment to be switched between types on the basis of patient response. 35