Fuzzy support for serodiagnosis: the ONSET program

ONSET derives its diagnosis from the combined evidence of contemporaneous and successive findings, which is in accordance with the behaviour of a clinician. Although it does not increase the diagnostic capabilities in principle, i.e., it cannot perform better than an expert reasoning on the basis of the same knowledge, computational power can be exploited to perform the matches and combine the results with greater precision, resolution, reliability, and transparency. Fundamental improvements of diagnosis, however, can only be expected from better serological tests or increased testing frequency. ONSET relies on domain knowledge explicated in a non-verbalized, analytical form. Knowledge representation-although mathematical-is intuitively clear and easy to conceive, making it equally accessible to both man and machine, and hence subject to objective evaluation and criticism.<>