Agent-Based User Interface Adaptivity in a Medical Decision Support System

Abstract

Previous work at Oxford Brookes University developed a system to advise on diabetes treatment that enabled data to be displayed according to the choices of each user. Due to the time critical nature of the problem, spending time searching through the data was not feasible. This reduced the usefulness of the system in the clinical setting for which it was designed. Thus a more automated approach was required. A multiagent system has been utilised to drive the adaptivity. A set of simple agents, each concerned with a single aspect of the system, communicate with each other and the suggested summary is a result of the emergent behaviour of the whole system. While emergent behaviour is used in other areas where agents have been applied, notably robotics, it is novel to use this approach in adaptive interfaces. This paper first considers the use of reactive agents to provide a context for the application of emergence in the area of self-adaptive interfaces. The field of adaptive interfaces is also considered to identify approaches that have been used in the past. An emergent multiagent system using a two-layer model is then described. This approach has been applied and tested to the problem of providing selfadaptivity at the interface to allow for decision support to be delivered in real-time for a clinician to employ