ABSTRACT TEMPORAL DIAGNOSIS IN MEDICAL DOMAINS

Most current model-based diagnosis formalisms and algorithms are defin ed only for static systems, which is often inadequate for medical reas oning. In this paper we describe a model-based framework plus algorith ms for diagnosing time-dependent systems where we can define qualitati ve temporal scenarios, Complex temporal behavior is described within a logical framework extended by qualitative temporal constraints. Abstr act observations aggregate from observations at time points to assumpt ions over time intervals, These concepts provide a very natural repres entation and make diagnosis independent of the number of actual observ ations and the temporal resolution. The concept of abstract temporal d iagnosis captures in a natural way the kind of indefinite temporal kno wledge which is frequently available in medical diagnoses. We use vira l hepatitis B (including a set of real hepatitis B data) to illustrate and evaluate our framework. The comparison of our results with the re sults of HEPAXPERT-I is promising. The diagnosis computed in our syste m is often more precise than the diagnosis in HEPAXPERT-I and we detec t inconsistent data sequences which cannot be detected in the latter s ystem. (C) 1997 Elsevier Science B.V.