Temporal Pattern Matching Using Fuzzy Templates

  • Authors:
  • Andrew Lowe;Richard W. Jones;Michael J. Harrison

  • Affiliations:
  • Dynamics and Control Group, Department of Mechanical Engineering, University of Auckland, Auckland, New Zealand. a.lowe@auckland.ac.nz;Department of Mechanical Engineering, University of Auckland, Auckland, New Zealand. r.jones@auckland.ac.nz;Department of Anaesthesia, Auckland Hospital, New Zealand. mikehar@ahsl.co.nz

  • Venue:
  • Journal of Intelligent Information Systems - Special issue on integrating artificial intelligene and database technologies
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

The identification of temporal patterns plays an importantrole in many medical diagnostic applications. Template systems thatidentify events and landmark points directly from time-seriesinformation have been shown to work well in various applications andin various forms. However, few such systems directly account for theuncertainty and vagueness often associated with medicaldecision-making. This paper describes a template system that usesfuzzy set theory to provide a consistent mechanism of accounting foruncertainty in the existence of events, as well as vagueness in theirstarting times and durations. Fuzzy set theory allows the creationof fuzzy templates from linguistic rules. The fuzzy template systemthat is introduced in this paper can accommodate multiple timesignals, relative or absolute trends, and obviates the need to alsodesign a regression formula for pattern matching (a requirement innon-fuzzy template systems)—the system automatically generates anormalised ‘goodness of fit’ score. Our target application for thefuzzy template system is anaesthesia monitoring. Initial testingusing both simulated and recorded patient data has been encouraging.Results are presented showing the diagnosis using various temporalrelationships of a number of problems.