Workflow technology has emerged as one of the leading technologies in modeling, redesigning, and executing business processes in several different application domains. Among them, the representation and management of health and clinical processes have been attracting a growing interest. Such processes are in general related to the way each health organization provides the required healthcare services. Health and clinical processes underlie the specification and application of clinical protocols, clinical guidelines, clinical pathways, and the most common clinical/administrative procedures. Current workflow systems are lacking in effective management of three general key aspects that are common (not only) in the clinical/health context: data dependencies, exception handling, and temporal constraints. For example, a laparoscopic intervention may need the results of the concurrent bioptic analysis to be properly concluded while exceptional recovery activities have to be performed in case of emergency evidence during standard treatment; however, the successful application of a fibrinolytic therapy requires a maximum delay of 30 min after the admission into the emergency department. In this paper, we propose TNest, a new advanced, structured, and highly modular workflow modeling language that allows one to easily express data dependencies and time constraints during process design, in addition to exception handling and compensation activities. As for temporal constraints, we focus here on temporal controllability which is the capability of executing a workflow for all possible durations of all tasks satisfying all temporal constraints. Moreover, we analyze the computational complexity of the temporal controllability problem in TNest, and we propose a general algorithm to check the controllability. All the features of TNest that have been considered to model clinical pathways from classical clinical guidelines, i.e., those features for the management of STEMI patients, published by the American College of Cardiology/American Heart Association, will be used throughout the paper as a motivating scenario.

Workflow technology has emerged as one of the leading technologies in modeling, redesigning, and executing business processes. The management of temporal aspects in the definition of a workflow process has been considered only recently in the literature. Currently available Workflow Management Systems ( WfMS ) and research prototypes offer a very limited support for the definition, detection, and management of temporal constraints over business processes. In this article, we propose a new advanced workflow conceptual model for expressing time constraints in business processes and we present a general technique to check different levels of temporal consistency for workflow schemata at process design time: since a time constraint can be satisfied in different ways, we propose a classification of temporal workflows according to the way time constraints are satisfied. Such classification can be used to successfully manage flexible workflows at runtime.

A Conditional Simple Temporal Network (CSTN) is a data structure for representing and reasoning about time-points and temporal constraints, some of which may apply only in certain scenarios. The scenarios in a CSTN are represented by conjunctions of propositional literals whose truth values are not known in advance, but instead are observed in real time, during execution. The most important property of a CSTN is whether it is dynamically consistent (DC), that is, whether there exists a strategy for executing its time-points such that all relevant constraints are guaranteed to be satisfied no matter which scenario is incrementally revealed during execution. Prior approaches to determining the dynamic consistency of CSTNs (a.k.a., solving the Conditional Simple Temporal Problem) are primarily of theoretical interest, they have not been realized in practical algorithms. This paper presents a sound-and-complete DC-checking algorithm for CSTNs that is based on the propagation of constraints labeled by propositions. The paper also presents an empirical evaluation of the new algorithm that demonstrates that it may be practical for a variety of applications. This is the first empirical evaluation of any DC-checking algorithm for CSTNs ever reported in the literature.