Advanced treatment of temporal phenomena in clinical guidelines

Advanced treatment of temporal phenomena in clinical guidelines 1

2

1

1

Paolo Terenziani , Luca Anselma , Alessio Bottrighi , Stefania Montani 1 DI, Univ. Piemonte Orientale “A. Avogadro”, Via Bellini 25\g, Alessandria, Italy 2 DI, Università di Torino, Corso Svizzera 184, 10149 Torino, Italy Abstract. Temporal constraints play a fundamental role in clinical guidelines. We sketch a computerbased temporal framework to represent temporal information in the guidelines, and to support different forms of inference and query-answering (which, e.g., might help in physician decision making). Context. Temporal indeterminacy, constraints about duration, delays between actions, and periodic repetitions of actions are essential in order to cope with clinical guidelines (e.g., with most therapies). Also the (implicit) temporal constraints derived from the hierarchical decomposition of actions into their components and from the control-flow of actions in the guideline must be considered. We sketch a temporal framework we are devising in order to extend GLARE, our computer-based and domain-independent manager of clinical guidelines [1], to represent temporal constraints in clinical guidelines and reason (i.e., perform inferences in the form of constraint propagation) with them. Methodology. We first propose a temporal representation formalism and constraint propagation algorithms operating on it, and then we show how they can be exploited to provide guideline systems with different temporal facilities. We devise a twolayer approach: (1) the high-level layer provides a high-level language to represent temporal phenomena and to offer several temporal reasoning facilities; (2) the low-level layer consists of an internal representation of the temporal constraints, on which temp oral constraint propagation algorithms operate. We designed our language in order to support tractable, correct and complete temporal reasoning. Results. In our high-level language, temporal constraints are represented by primitives such as “delay(X,Y,min,max)” stating that the delay between actions X and Y is between min (minimum delay) and max, “duration(X,min,max)” and so on. More complex primitives are used to represent repeated/periodic actions. At the low level, all constraints are mapped onto an STP-Tree [2]. Our approach provides the following “temporal reasoning” facilities: 1. the consistency-checking-guideline facility, to check the temporal consistency of the guideline. It can be advocated at any stage during the

acquisition of a clinical guideline, so that incremental consistency checking is also possible. 2. the consistency-checking -instance facility, to check whether the temporal constraints in the guideline have been respected by the instances of actions executed on specific patients. 3. the query facilities, to grasp temporal information, e.g., as an help in decision making. These set of facilities includes: (i) the next-action facility, to assess when the next actions have to be performed, given the constraints in the whole guideline and given the time when the last actions have been executed. (ii) the yes/no query facility, to ask whether a given set of temporal constraints is possible given the set of temporal constraints in the guideline. (iii) the extract facility, to output the temporal constraints between a given set of actions. (iv) the hypothetical query facility, to ask queries in the hypothesis that some temporal constraints is assumed (e.g., “If I perform action A1 at 12:30, when will I have to perform A2, …,An?”). Recently, we have also devised advanced Temporal Databases techniques to both store and query properly temporal data in a relational framework [3]. Conclusions . Our approach shows that by extending and applying advanced Artificial Intelligence techniques for temporal reasoning, guidelines systems can be provided with a powerful framework to manage different temporal phenomena, greatly improving the effectiveness and usefulness of clinical guidelines themselves. References [1] P. Terenziani, G. Molino, M. Torchio. A Modular Approach for Representing and Executing Clinical Guidelines. Artificial Intelligence in Medicine Journal 23 (2001) 249-276. [2] L. Anselma, P. Terenziani, S. Montani, A. Bottrighi. Towards a Comprehensive Treatment of Repetitions, Periodicity and Temporal Constraints in Clinical Guidelines. Artificial Intelligence in Medicine Journal, to appear. [3] P. Terenziani, R.T. Snodgrass, A. Bottrighi, M. Torchio, G. Molino. Extending temporal Databases to deal with telic \atelic medical data. Artificial Intelligence in Medicine Journal, to appear.

AMIA 2006 Symposium Proceedings Page - 1117