Many projects are under way to create or derive computational models of biological processes, using one or more biomedical ontologies or other biomedical knowledge resources as a basis for the model structure. One such project, which we have started, is to model the local/regional spread of tumor cells (metastasis) so that an accurate target for radiation therapy can be defined. This approach uses the structure of the lymphatic system as defined in the UW Foundational Model of Anatomy (FMA) to produce a tumor dissemination model for any specific anatomic site where the primary tumor may be located. Such models are tedious to build manually and are prone to errors in transcription of the structures. We propose to further develop this approach to modeling tumor dissemination as a case study for developing automated methods for generating dynamic models from biomedical ontologies. The project will investigate the use of new mathematical formalism as a way to facilitate the mapping from an ontology to a model, and apply the formalism to the FMA/tumor dissemination problem as a proof of concept. The specific medical problem we are addressing has the potential to significantly increase the accuracy of radiation therapy target volume definition. This is a criticl step in more fully utilizing the power of modern computer controlled radiation therapy treatment machinery and techniques. The general methods may significantly help researchers to generate new models, insure that such models are consistent, and also possibly aid in identifying any inconsistencies in the biomedical ontologies themselves.
Much biomedical knowledege is now in a form that computer programs can read and use. This project will develop powerful new mathematical ideas that can help make writing these programs much easier. The specific problem of designing radiation treatment for cancer will be a pilot project, and if successful, the methods should be useable for computational modeling of public health knowledge as well as medicine and basic biology.