Mike Sharples, Nathan Jeffery, Derek Teather, Briony Teather, George du Boulay
Tutoring systems could satisfy a demand of many professions for structured case-based training, but to be accepted they need to be robust, authoritative and matched to the needs of trainees in the workplace. This paper outlines a methodology for the development of knowledge-based training that integrates software, task, knowledge and organizational engineering. It consists of a set of 'building-blocks' that specify the type of activities needed to develop a complete knowledge-based training system, while allowing flexibility in the choice and ordering of specific design techniques. The approach is illustrated by a project to develop the MR Tutor, a knowledge-based training system for neuroradiology. The building-blocks for this project have included an analysis of published studies of cognitive processes in medical image interpretation, elicitation and refinement of knowledge from an expert neuroradiologist, workplace studies of radiology training and experiments with new techniques for data visualization. The MR Tutor gives trainee radiologists the experience of observing and analysing a large archive of cases and practice in comparing their interpretations with those of experts. It is based on a structured language for describing abnormal appearance in Magnetic Resonance images of the head and it uses a novel 'overview plot' to visualize and interact with the image archive. The development methodology has been followed to the stage of implementing a robust integrated system.
Download compressed postscript file