The advent digital health information in the form of electronic health records, clinical information systems and online biomedical evidence is leading to many decision sciences researchers to look at methodologies for developing clinical decision support systems. A working definition of clinical decision support systems has been proposed by the Centre for Health Evidence as; "Clinical Decision Support systems link health observations with health knowledge to influence health choices by clinicians for improved health care".
In this talk I will outline the design and development of two component parts of a distributed clinical decision support system where the current application domain is focused on paediatric asthma. The first component is an intelligent prediction model that, given a representation of a current asthma patient in terms of clinical attributes and values tries to predict the severity of an asthma exacerbation. I will describe the design of the model in terms of clinical specifications, classification methodologies, and present a set of preliminary results from a retrospectively collected asthma dataset.
The second model I will present provides a methodology for retrieving clinical evidence from an online biomedical repository so that it may be used for point of care decision making. The development of the model consists of two main tasks; the creation of an ontology for representing relevant clinical concepts and for resolving interoperability issues between clinical terms used by local clinical decision support systems and index terms used by repositories of evidence-based documents, and the development of an information retrieval engine for leveraging evidence from online biomedical repositories that can support patient-specific decision making. An initial set of results for the clinical evidence model will be presented, and both components are currently being evaluated as part of a larger live clinical trial of the complete distributed decision support system.