Introduction The increasing complexity of cancer diagnostics and more individualized treatment options, also in head and neck oncology, require new patient information processing techniques and decision support systems in the head and neck tumor board (HN-TB). Therefore, a digital patient model (DPM) for laryngeal carcinoma (LC) was developed on the basis of Bayesian networks (BN) and positively evaluated.
Methods Now that the LC model has been successfully developed, oropharyngeal carcinoma (OC) as another entity was modeled as a BN graph . The OC model was created according to accepted guidelines and analyzes of HN-TBs at the University of Leipzig. The graph structure was optimized and compared with the LC model.
Results The oropharynx model contains over 250 information entities connected by about 350 edges. A particular challenge was the implementation of the 8th edition of the TNM classification in the model that highlights the HPV/p16 status, which made model construction more difficult compared to the LC model. An expert-based evaluation and optimization of the model structure was conducted.
Conclusions Personalized medicine and targeted therapy are increasingly important in oncologic treatment and require a structured and comprehensive support of information management and decision making. The BN model of the OC is currently in construction. The graph structure was created and optimized. After validation, the model should be tested in order to support the therapy decision processes in the OC perspectively.
Poster-PDF
A-1789.PDF
