Author:
Monique Hendriks
Affiliation:
Philips Research, Netherlands
Keyword(s):
Clinical Decision Support Tools, Clinical Prediction Modeling, Inclusion of Domain Knowledge, User Interface Design, User Evaluation, Data Visualization.
Related
Ontology
Subjects/Areas/Topics:
Artificial Intelligence
;
Biomedical Engineering
;
Business Analytics
;
Cardiovascular Technologies
;
Computing and Telecommunications in Cardiology
;
Data Engineering
;
Data Mining
;
Databases and Information Systems Integration
;
Datamining
;
Decision Support Systems
;
Decision Support Systems, Remote Data Analysis
;
Enterprise Information Systems
;
Health Engineering and Technology Applications
;
Health Information Systems
;
Information Systems Analysis and Specification
;
Knowledge Management
;
Knowledge-Based Systems
;
Ontologies and the Semantic Web
;
Practice-based Research Methods for Healthcare IT
;
Sensor Networks
;
Signal Processing
;
Society, e-Business and e-Government
;
Soft Computing
;
Symbolic Systems
;
Web Information Systems and Technologies
Abstract:
As healthcare is becoming more personalized, prediction models have become an important tool for decision support. In order to create sensible, understandable and useful prediction models, it is often necessary to include domain knowledge. This requires multi-disciplinary communication which has proven to be difficult, as the different parties involved are not always aware of each other’s information needs. This paper presents the design process of a tool which supports the communication between clinical experts and data mining experts. Interviews and user tests were executed on four different sites and with 14 different users from both domains. The results from these user tests confirm the need for support on the communication process and provide evidence that the tool presented here indeed provides support by helping both parties to understand each other’s information needs. The tool provides a graphical user interface which guides the users through the steps required to create a p
rediction model. The graphical user interface helps the clinical expert to understand the choices to be made which rely on his/her expertise, while the fact that a ‘quick-and-dirty’ first version of a prediction model is generated in the process, helps the data mining expert to uncover all formal requirements for the model.
(More)