gms | German Medical Science

62. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

17.09. - 21.09.2017, Oldenburg

Evaluation of an Ontology-based System for Risk Minimization in Hospitals

Meeting Abstract

Search Medline for

  • Kais Tahar - IMISE/Universität Leipzig, Leipzig, Deutschland
  • Alexandr Uciteli - IMISE/Universität Leipzig, Leipzig, Deutschland
  • Heinrich Herre - IMISE/Universität Leipzig, Leipzig, Deutschland

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 62. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Oldenburg, 17.-21.09.2017. Düsseldorf: German Medical Science GMS Publishing House; 2017. DocAbstr. 142

doi: 10.3205/17gmds111, urn:nbn:de:0183-17gmds1112

Published: August 29, 2017

© 2017 Tahar et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution 4.0 License. See license information at http://creativecommons.org/licenses/by/4.0/.


Outline

Text

Introduction: As the strain of medical staff and complexity of patient care grow, the risk of inducing medical errors only increases. However, between 50 and 60% of adverse events on medical decision-making could be avoided through better organisation and more attention [1]. In the research project OntoMedRisk (BMBF grant no. 01IS14022), we developed an ontology-based system for risk detection and analysis in hospitals [2]. This paper focuses on the evaluation of our ontology-based software module (OSM).

Methods: The OSM developed consists of four components: the Risk Identification Ontology (RIO), a Risk Specification Template (RST), the Risk Ontology Generator (RIOGen), and the Ontology-Based Risk Detector (OntoRiDe). RST allows experts to specify risk classes including their corresponding risk specification rules (RSRs). RIOGen generates RIO entities from RST. For example, the infection risk was transformed into a subclass of (is-a) RIO:Risk using RIOGen. In addition, the RSR of infection risk was encoded as an anonymous equivalent class in OWL. Based on RIO, OntoRiDe enables the identification of perioperative risks. This tool receives the properties of potential risk situations called KPIs (Key Performance Indicators) as input parameters and checks all RSRs in order to detect the specified risks in the current situation.

To test these components, we implemented a JUnit test suite (JTS) and chose the Cochlear Implantation (CI) as the first exemplary clinical treatment process. First, risks related to CI were specified in RST and automatically transformed in RIO using RIOGen. This process was analyzed by several JUnit tests that analyze and count the generated ontological entities. These were then compared with the number, names and types of the predefined entities in RST. Second, we classified the CI risks based on their RSRs in four types with different degrees of complexity: simple risks are represented by simple RSRs defined by a single KPI, disjunctive risks are represented by composite RSRs defined by a union (OR operation) of several KPIs, conjunctive risks are represented by composite RSRs defined by an intersection (AND operation) of KPIs, and finally, hybrid risks are represented by composite RSRs defined by a combination of conjunctions and disjunctions of KPIs. In addition, we used spreadsheets as a source of input data to test each risk type using at least two examplary KPI sets. The first set of KPIs fulfills the corresponding RSR, while the second set does not fulfill this rule. The JTS developed initializes OntoRide using these KPI samples and compares the detected risks with those that were expected.

Results: The OSM components were successfully tested and evaluated using the JTS developed. The test results have indicated the correctness of the generated ontological entities. OntoRiDe was also tested and evaluated through 8 KPI sets that represent all possible types of perioperative risks. The test results reveal the correctness of the risk detection method.

Discussion: The system developed yields promising results, which can be used to alert medical staff and avoid perioperative risks. Our future studies will integrate this system into the existing IT infrastructure at Jena University Hospital.



Die Autoren geben an, dass kein Interessenkonflikt besteht.

Die Autoren geben an, dass kein Ethikvotum erforderlich ist.


References

1.
Kohn LT, ed. To Err Is Human: Building a Safer Health System. 7. print. Washington, DC: National Acad. Press; 2008.
2.
Uciteli A, Neumann J, Tahar K, Saleh K, Stucke S, Faulbrück-Röhr S, et al. Risk Identification Ontology (RIO): An ontology for specification and identification of perioperative risks. In: 7th Workshop on Ontologies and Data in Life Sciences; Halle (Saale), Germany; 2016.