Using a Knowledge-based System for Validating and Classifying Requests in Large-scale Laboratories, Considering Preliminary Diagnoses.

Motivation: In general, an expert system for improving quality assurance for incoming requests in large-scale laboratories is designed and implemented, which works to a large extend independent from human individuals as it uses past request data supplied by the laboratories. In particular, this expert system (LabExpert) is extended to incorporate not only information on valid test compositions for requests, but also information on preliminary diagnoses using NLP (natural language processing) methods to enable the underlying knowledge-base to perform a more precise quality assessment. Results: The incorporation of preliminary diagnoses to LabExpert only provides significant results within a given gray area, as the number of requests containing an utilizable preliminary diagnosis within the training data account for only 6.7%. Within this area, LabExpert shows clearly the benefit of using NLP methods to provide further parameters for classification and adaptation. Availability: LabExpert is not available due to special nondisclosure agreements with cooperation partner.