Fuzzy logic in diagnostics of rare diseases

Abstract

Background. Rare disease (RD) is any disease that affects a small percentage of the population, with a prevalence of about 1 in every 2000 people. Diagnostic of RDs is hindered by the lack of knowledge predetermined by the multitude of these diseases, which can lead to medical mistake and even medical failure. It may also occasion the omission of the necessary investigations and vice-versa the prescription of a multitude of unnecessary and potentially hazardous invasive diagnostic interventions and/or delay in performing them. Quite frequently the correct diagnosis is belated, sometimes it is not made at all. The generally accepted way for RD diagnosing is usage of search machines, but this way is reasonable only when some outstanding symptoms/signs, like dismorphological signs, mental retardation, etc. occur. Very often the RDs mask as common diseases. In such cases the problem is to suspect the RD. There are no special algorithms which give rise to a suspicion for the RD. Objective. Elaboration of the model/algorithmfor revealing cases suspicious of RDs (if they present under the mask of a common disease). Methods. We need to have an approach which will help us to suspect the RD to use the search machines afterwards as usually. We assume to suspect a RD when the clinical picture and course of a disease are atypical. But the point is that the border between typical (normal) and atypical (abnormal) is not crisp: the fuzzy methodologies can be used here. Results. We present an algorithmic approach for the implementation within a framework of a computer program, that would allow to suspect RD, and therefore serve as a basis of a decision support system. Examples from medical practice illustrate our approach. Conclusions. For the group of common diseases (syndromes), e.g. pneumonia, bronchitis, rheumatic fever, etc., we propose to prepare medical electronic records (including complains, anamnesis aegroti/vitae, anamnesismorbi, status presence, etc.), which would reveal the signs/symptoms which deviate from the normal clinical course (by frequency, intensity, time of manifestation, duration, etc.). When deviation reaches some level, the program would signal that there is some suspicion of the non-common disease (e.g., RD) which masks as a common one. © Springer-Verlag Berlin Heidelberg 2013.