You are here: Home Medical Informatics Projects Project Debug IT

Project Debug IT

Title: DebugIT: Detecting and Eliminating Bacteria UsinG Information Technology. A Large-scale integrating project

Active Period: 2008 until December 2011

Primary investigator for project management: Jose Verguts, Dirk Coalert (AGFA Healthcare)

Primary investigator for ontology work package: Daniel Schober (UKLFR-IMBI)

Freiburg team: UKLFR-IMBI: Daniel Schober, Martin Boeker, Ilinca Tudose, Maren Kechel , Averbis GmbH

Funding: EU 7th Framework program (7th FP)

Research Question

The DebugIT ontoloy workpackage, which the UKLFR team is leading, develops the 'DebugIT Core Ontology'(DCO).
This OWL DL ontology is the key mediator for the semantic and syntactic data communication and integration between the participants within the DebugIT interoperability plattform. DCO  ensures that heterogenous data content from various sources can be mapped on an accepted, universally reuseable common vocabulary with computer interpreteable semantics. Such DCO annotated data can be queried uniformly via semantic web query languages, easing cross site data integration. The following has to be ensured:
a) term coverage: all terms as used in querying need to have a counterpart construct in DCO.
b) the exchange syntax can be read and processed by many software tools (standardized syntax in RDF-OWL)
c) the grammatical query complexity needs to be coped with by the semantics of the representation language (OWL-DL plus rules).
d) Choosen semantics and syntax allow for evaluation via semantic web technologies such as rule-based and logical reasoning, consistency checks, as well as automatic inferrence of new knowledge from given facts.
Further questions are: Where can the ontologies be simplified by a) removeing complexity, or b) hiding complexity from the user?
Can the gap between the content variability of the clinical data be closed by automatic mappings that bridge to ontological concepts, ensuring a minimum amount of queries can be answered in satisfactory granularity ? (An example question would be: "What percentage of Escherichia coli cases, cultured from urine samples, is resistant to the combination of trimethoprim / sulfamethoxazole (TMP / SMX) or trimethoprim over a period x?")
Are the few well-established terminologies in use sufficient in order to have enough data mapped to ontology concepts, so that the competency questions can be answered in sufficient granularity ?
Can clinicians and non-ontologists formulate nontrivial ontological questions and understand the results?
Can parts of the ontology be serialised into constraint natural languages and therefore be submitted to non-ontologists for evaluation?
How can logic reasoning by used during ontology engineering, for evaluation and usage of ontology ?


The project aims to create a development platform for high-throughput analysis of distributed clinical data with regard to the spread of antibiotic resistance of infectious pathogens in European hospitals. In an iterative approach, data is to be retrieved , represented homogeneously via ontologies. The unified data is then analysed using rule-based systems to detect trends and risk patterns. This analysis provides new knowledge, which in turn can lead to 'Patient Safety Guidelines', 'decision support' and 'Early-Warning Systems'.

Short summary of project

The DebugIT System ("Detecting and Eliminating Bacteria Using Information Technology") is used to prevent the spread of antibiotic resistance in European hospitals. Clinical data from 7 different locations are made available and compareable via .  data warehousing and semantic web technologies which integrate access and semantics uniforml. The formalized result data sets are amenable for intelligent data mining approaches illustrating and analysing the differences in  drug prescribtion practices and their effects on the spread of antibiotics resistance. DebugIT implements an interoperability platform using communication standards such as ontologies and SPARQL data content exchange and processing. The system follows a linked data approach.