Towards an automated assistant for clinical investigations

Before a drug can be made available to the general public, its effectiveness has to be experimentally evaluated. Experiments that involve human subjects are called Clinical Investigations (CIs). Since human subjects are involved, procedures for CIs are elaborated so that data required for validating the drug can be collected while ensuring the safety of subjects. Moreover, CIs are heavily regulated by public agencies, such as the Food and Drug Administration (FDA). Violations of regulations or deviations from procedures should be avoided as they may incur heavy penalties and more importantly may compromise the health of subjects. However, CIs are prone to human error, since CIs are carried out by the study team, which might be overloaded with other tasks, such as hospital and/or pharmacy duties, other trials, etc. In order to avoid discrepancies, we propose developing an automated assistant for helping all the parties to correctly carry out CIs as well as to detect and prevent discrepancies as early as possible. This way the proposed automated assistant would minimize error, and therefore increase the safety of the involved subjects. This paper takes the first steps towards that direction. In particular, we propose a model for collaborative systems with explicit time, called Timed Local State Transition Systems (TLSTS), and argue that it can be used for specifying procedures and regulations for CIs, which mention time explicitly. Finally we show how to implement a TLSTS specification using Maude, an existing computational tool based on rewriting.