Knowledge Discovery and Data Engineering Research Group


  inFormation Extractor of cLInical Practice guidElines




A short description

FELIPE is a system which, in completely automatic mode, extracts and structures information of Clinical Practice Guidelines (CPGs) from textual documents. Original guidelines are unstructured or semi-structured textual documents (e.g., guidelines delivered by National Guideline Clearinghouse) Each CPG contains practices and indications of a specific health condition/disease for a set of medical decision problems (e.g., diagnosis, treatment, etc.).The output generated by FELIPE is a filled template in the form of a XML-file compliant the standard Clinical Document Architecture (CDA-HL7 v3): each final guideline contains practices and indications concerning a specific health condition/disease for a specific medical decision problem, while each outer node contains a specific practice or indication of the original CPG.

Top of this page  

Functional Architecture

The functional architecture consists of two main components:

  • Text-Analysis component which includes Document Structure Extraction, Text-Processing/Tagging blocks of Figure 1. It localizes, recognizes and annotates relevant pieces of information from the sections composing CPGs by exploiting linguistic resources and the natural language processing functionalities available in GATE.
  • Guideline Structuring component which includes Template Filling, Enrichment, Output Generator blocks of figure 1. It fills a template structure with relevant pieces of information extracted by Text-Analysis component. The filling procedure follows a knowledge-based approach (production system in CLIPS) based on knowledge engineer and domain expert. The rule base is defined in terms of the annotations generated by Text-Analysis component.

Figure 1: Functional architecture of FELIPE system.

Top of this page

Web Server Architecture

FELIPE was also integrated into a web-server framework (Figure 2). The process of extraction of the information of CPGs is completely executed on side-server, then, once structured in XML files, guidelines are stored in a relational database through a transformation step (DB Populator) from semi-structured representation to the structured relational one. Clinical Practices Guidelines are then accessible and retrievable by means of two web-services, namely Query Web Service Request Guideline Web Service, invokable through a web-based Interface. The first retrieves a result set of possible CPGs, while the second returns an only one and specific CPG. The results of both queries are represented in the standard CDA-HL7 v3 formalism.


Figure 2: Block-diagram of the web-server architecture hosting FELIPE system.

Top of this page

Metabolic Syndrome Application

This application consists of the extraction and structuring of information contained in some sections of the textual CPGs concerning metabolic pathologies and syndromes. In this application FELIPE exploits domain-dependent controlled vocabularies and grammars as well as a rule-base hand-coded by knowledge engineers and domain experts ().Consider, for instance, the guideline of Metabolic Syndrome which illustrates practices and indications for the decision problems of Diagnosis, Management, Prevention, Risk Assessment, Treatment (see Figure 3)


Figure 3: An excerpt of the clinical guideline on "Metabolic Syndrome".

The information extracted and structured by FELIPE the problem of Diagnosis is displayable follows:


Top of this page

Project team

      Project Leader

Donato Malerba

      LACAM Staff

Corrado Loglisci Michelangelo Ceci

    Students involved in the project Galiano Francesco, Minervini Pasquale, Palmisano Tonino


Related publications

(in inverse chronological order)

  • C. Loglisci, M. Ceci, D. Malerba. A Knowledge-Based Framework for Information Extraction from Clinical Practice Guidelines. International Symposium On Methodologies For Intelligent Systems, ISMIS 2009. J. Rauch et al. (Eds.): , LNAI 5722, pp. 119–128, 2009. Springer-Verlag (Germany)

Top of this page

Last Update: Mon Aug 10 2015 16:39:48 GMT+0200 (CEST)Last Update: Wed Jun 13 2007 05:13:33 GMT+0200 (ora legale Europa occidentale)

KDDE  Template

KDDE presentations have to be based on this template.

Group members and students who are taking a degree, are invited to use it.

Discovery Science 2016

The 19th International Conference on Discovery Science (DS 2016) will be held in Bari on October 2016, 19th-21st. KDDE Group is organizing it.

ALT 2016

Algorithmic Learning Theory 2016

Bari, Italy, 19-21 October, 2016.

Powered by CMSimple| Template:| Login