Medical Cause and Effects Analysis (MCEA

In the MCEA project, a novel hybrid approach to knowledge modeling for medical expert systems is being developed. On the one hand, established "Cause and Effect Analysis" methods in other domains (e.g. automotive) including a powerful software platform are used for a declarative representation of mainly causal knowledge (e.g. symptom-disease). On the other hand, this represented knowledge forms the basis for machine learning methods based on it for prognosis, diagnostics and therapy within selected clinical use cases.

MCEA is being worked on by six computer science institutes of the University of Lübeck and six clinics of the University Hospital Schleswig-Holstein (UKSH) under the coordination of the UniTransferKlinik and with the participation of the company PLATO AG and is being funded within the framework of the AI strategy of the state of Schleswig-Holstein.

The focus of the subproject worked on by IMI is a MCEA-specific utilization of SNOMED CT as the most internationally expressive terminology. First, a broad annotation of used concepts in MCEA models ensures interoperability of the modeled machine knowledge, e.g. for the application of the knowledge to available, often differently represented patient data (see "Curly brace problem", LINK) or for links to further external knowledge like drug databases or guidelines. Second, valuable knowledge represented via the formal-logically represented reference terminology SNOMED CT is itself of interest. Due to the complexity of SNOMED CT with its approx. 350,000 concepts (LINK), partial sections of the concept system are of interest (e.g. of anatomy, physiology, ...), which can be made available via an interface for specific MCEA knowledge models.

Project team

Prof. Dr. H. Handels
Prof. Dr. J. Ingenerf (Subproject lead)
M.Sc. A. Banach
M.Sc. S. Germer