Providing embedded proactive task support for diagnostic jobs: a neural network-based approach

摘要

This research proposes the use of a neural network (NN) to provide proactive task support, which is embedded in user's primary application for diagnostic jobs. What is unique about this research is that the NN serves as not only the backend inference engine, but also the driver of the user interface at the front-end. The user interface features separate and persistent advice windows, where procedural advice and relevant domain knowledge are displayed continuously, side-by-side to the task window. The display is updated at short intervals, according to the task progress. At the backend, the NN, complemented by a simple semantic network, makes inference about a user's task and continuously refines its inference. A prototype has been built to demonstrate the proposed approach. It supports novice nurses’ triage task based on a telephone interview. The usability of the prototype has been informally evaluated at a medical call center and the feedback was largely positive.