Ontology-based student testing through clinical guidelines: An AI approach

Abstract

On the basis of our 25-year experience with the GLARE (Guideline Acquisition, Representation and Execution) clinical decision support system, we have started to analyze the adoption of computer-interpretable clinical guidelines (CIGs) and AI techniques to train and test medical students about how to act on patients . Moving from decision support to the educational task involves significant research challenges. In this paper, we propose a new facility that supports teachers in the definition of tests, by selecting and hiding to students specific parts of the CIG, and asking students how they would act on the given case study (patient) in the selected parts. Students are provided with a medical ontology to identify proper actions/decisions, and students' proposals are then automatically compared with what the CIG (considered as a “golden standard”) would suggest to do to the patient through knowledge representation and reasoning techniques. Our basic explanation mechanism exploits the medical ontology to show to students the differences (if any) between their proposals and the ones of the CIG. • New educational approach based on Computer-interpretable clinical guidelines (CIGs) • Training and testing medical students about how to act on patients (following CIGs) • Medical ontology for action selection, conformance check (wrt CIG), and explanation • Test definition and acquisition

Ähnliche Arbeiten