Public health alert system for health networks: application to cardiology

The authors propose a system that allows healthcare professionals like physicians and nurses to define medical alerts from patient and environmental data by using fuzzy linguistic variables. Such variables are associated to three importance levels (very important, important or less important) indicating their relative importance in the context and can be developed separately from alerts. Each time a predefined alert is activated by the system, it has two quality indicators which are used for filtering: an 0 to 1 applicability level stating how much the patient is concerned and a trust level indicating its reliability and calculated according to the amount of information that is available at the moment. Finally, lack of information, very common in medical records, is treated transparently thanks to the new concept of modifier, which allows to express the influence variables have on each other by means of a weighted oriented graph