Computer-assisted clinical diagnosis in the official European union languages

Abstract

eHealth services integrate Web Information Retrieval and Intelligent Medical Decision Support for health care professionals based on the range of possible symptoms which a patient reports. However, many symptoms like high temperature, fever, or headache, are ambiguous in terms of suggesting wide variety of possible patient's conditions to the GP, while other symptoms are mutually dependant, which again can be misleading to make an accurate diagnosis. On the other hand, doctor's up-to-date knowledge on the medicaments, drugs, active medical substances included, anticipated range of diseases relating to the symptoms reported, and the most reliable pharmaceutical manufacturers, are of the greatest importance to cure the illness successfully. This study proposes an approach to support so called standard medical procedure or clinical guidelines in treatment of each of the diseases by delivering such a knowledge to the physician and by individualizing the selection of drugs in respect to the patient's specific needs in order to avoid a potential drug interaction. We evaluate efficiency of a medical multilingual decision support system Diagnosia on the grounds of accessibility to such a knowledge depending on the EU language and we use Bayesian inference for generating the optimal decision on reaching a particular diagnosis accuracy. Our methodology is examined on the real data taken from the American service Prescriber Checkup and some other knowledge-based medical resources of nationally recognized rank. Our findings indicate that this approach outperforms not only traditional standard procedure of curing some commonly occurring illnesses, but also many commercial computer-assisted medical support diagnostic systems.