Knowledge-based generation of diagnostic hypotheses and therapy recommendations for toxoplasma infections in pregnancy.

ABSTRACT

Primary infection of pregnant women with the parasite Toxoplasma gondii results in infections of the unborn by transplacental transmission in about 50% of the cases. The degree of possible damage depends on the duration of parasitical impact on fetal tissues. The web-based software system ToxoNet processes the results of serological antibody tests performed during pregnancy by means of a knowledge base containing medical knowledge on the interpretation of toxoplasmosis serology findings. For this purpose, it matches the results of all serological investigations of maternal blood with the content of the knowledge base and generates interpretive reports consisting of a diagnostic hypothesis, recommendations for therapy, and proposals for further investigations. Fuzzy sets are used to formalize certain intervals between subsequent investigations to take the varying immune responses of individual patients into account. In a retrospective study, ToxoNet classified 100% of the trivial serological cases and about 87.8% of the more complex cases correctly. ToxoNet comprises a knowledge base, a system for interpretation, and a knowledge acquisition and modification program. It is available on the WWW by accessing a medical knowledge-base server via standard browsers.