HEARTFAID Decision Support System Prototype

Colantonio S., Martinelli M., Moroni D., Salvetti O., Chiarugi F., Tsamardinos I., Candelieri A., Conforti D., Lagani V., Gamberger D.

Decision Support System  Prototype 

This deliverable summaries the activities carried out, within Task 5.4 "Implementation of the Decision Support System", for developing the HEARTFAID CDSS Prototype, following the design that was performed in the previous years of the Project, within Task 5.3, and summarized in Deliverable D15 "Functional Specifications of Data Processing and Decision Support Services" (DEL D15, 2007). Such a design was based on a deep investigation of CHF stakeholders' needs and expectations which resulted in a detailed list of CDSS functional requirements; while an accurate analysis of the methodological and technological foundations resulted in the detailed definition of the CDSS functional specifications. The CDSS has been then devised by integrating, in functionally advanced settings: (i) deductive knowledge, elicited from guidelines and medical experts; (ii) inductive knowledge, extracted by data mining techniques applied to significant sets of data; (iii) computational methods for the analysis and interpretation of diagnostic data (i.e., ECG signals and Echocardiographic images). This has required the collection of some results of the activities carried out in other Tasks or Work Packages (WP) of the project. In particular, the deductive knowledge base has been formalized in WP4 and a precise description of its content can be found in Deliverable D22 "Ontologies and Knowledge Representation" (DEL D22, 2007). Inductive models have resulted from the activity of knowledge discovery and described more precisely in Deliverable D29. Methods and algorithms for processing and analysing diagnostic ECG and Echo images have been developed in Task 5.2 and detailed in Deliverable D30 "Models and Methods for Signals and Images Processing" (DEL D30, 2008). This document mainly focuses on the implementation of each component of the CDSS architecture and their integration into the system, and on the integration of the CDSS itself within the platform (integration that is realized mainly by means of the HEARTFAID Web Portal). For the implementation activity, Semantic Web Technologies (SWT) have been used as the most advanced tools for formalizing, re-using and sharing medical knowledge, and reasoning on it; while a service oriented approach has been adopted for the integration and easy access to a number of functionalities. Moreover, realistic clinical scenarios have been carefully defined in cooperation with clinical partners for assuring the realness and effectiveness of the developed functionalities of the system. More in details, the document is organized as follows. In the first introductory chapter, a recall of the significance of decision support in the management of CHF is reported along with a brief description of the main issues to be faced when developing a decision support system. Afterwards, in the second chapter, HEARTFAID CDSS architecture, as was defined in the previous Tasks 5.3 of the Work-Package 5, is described and motivated. Also, the integration of the system within HEARTFAID platform is summarized for assuring a global view of the system functioning. In Chapter 3, attention is focused on the implementation details, hence each component of the CDSS architecture is considered and its implementation described in details along with the problems faced and the solutions adopted. Particular attention is devoted to the ontological Knowledge Base, the computational reasoning methods included into the Model Base, as well as the algorithms defined and developed for processing ECG signals and Echocardiographic images. Finally, the implementation of the Meta Level of the system is described and discussed. In the fourth chapter, the integration within the Platform through the Middleware is discussed along with the issues faced and the technological choices made. The implementation of the showcase related to the clinical management of a HF patient is reported in Chapter 5. Two appendices close the document.

Source: Project report, HEARTFAID, Deliverable D36, 2008