Event calculus to support temporal reasoning in a clinical domain

This work concerns temporal aspects of a knowledge based system which holds information on patients as they progress through their treatment in a vascular surgery department. Representing and using knowledge about temporal relationships so as to provide decision support to a historical knowledge base of patient data is investigated. Event Calculus, in first order classical logic augmented with negation by failure, provides an effective framework for reasoning about time. From Kowalski and Sergot's original Event Calculus we arrive at a simple and flexible framework which can be used as a temporal support in a medical knowledge based system. We show how Event Calculus can be used to describe a simple model of the clinical pathway in vascular surgery. Patient information in the medical record is formalised in a structural framework to suit the Event Calculus. Medical knowledge about investigation and treatment options is added to the model so that the resulting system can recommend the options which are appropriate at any particular time. It is shown how these recommendations provide decision support by recommending what should be done next, and when to re-evaluate measurements that become unreliable. It is argued that there are advantages to be gained by adopting a general temporal reasoning framework because it can be extended to support various medical and administrative tasks. The extensions available to the Event Calculus, further its suitability as a temporal reasoning framework in the medical domain. A prototype system, essentially a research workbench over a realistic domain, is built using Prolog to illustrate the temporal reasoning capabilities provided by the Event Calculus framework. Using case studies it is demonstrated how the prototype system fulfils the decision support abilities we aimed to achieve in the knowledge base.