Abstraction and Analysis of Clinical Guidance Trees

Abstract

OBJECTIVES The aims of this work were: to define an abstract notation for interactive decision trees; to formally analyse exploration errors in such trees through automated translation to LOTOS (Language Of Temporal Ordering Specification): to generate tree implementations through automated translation for an existing tree viewer, and to demonstrate the approach on healthcare examples created by the CGT (Clinical Guidance Tree) project. APPROACH An abstract and machine-readable notation was developed for describing Clinical Guidance Trees: ADIT (Abstract Decision/Interactive Trees). A methodology has been designed for creating trees using ADIT. In particular, tree structure is separated from tree content. Tree structure and flow are designed and evaluated before committing to detailed content of the tree. Software tools have been created to translate ADIT tree descriptions into LOTOS and into CGT Viewer format. These representations support formal analysis and interactive exploration of decision trees. Through automated conversion of existing CGT trees, realistic healthcare applications have been used to validate the approach. RESULTS All key objectives of the work have been achieved. An abstract notation has been created for decision trees, and is supported by automated translation and analysis. Although healthcare applications have been the main focus to date, the approach is generic and of value in almost any domain where decision trees are useful.