Specifying and representing temporal abstractions on clinical data by a query language based on the event calculus

Presents a system for temporal abstraction of clinical data. This system was developed following an object-oriented approach, and adopting the event calculus (EC) as a temporal ontology. EC is a general theory of time and change based on the concepts of event and property: an event happens at a time point and initiates/terminates one (or more) properties, while a property holds during a time interval. Examples of abstract properties in our domain are parameter state (normal, high, low), and parameter trend (increasing, decreasing and stationary). The system was tested by evaluating time-stamped sequences of numeric echocardiographic parameters (e.g. ventricular diameters, Doppler flow velocities, valve-lumen areas, etc.) in different clinical conditions (e.g. myocardial infarction, valve disease, etc.). A suitable graphical representation of the performed temporal abstractions is also provided by the system.